laser desorption/ionization mass spectrometry for direct
TRANSCRIPT
Retrospective Theses and Dissertations Iowa State University Capstones, Theses andDissertations
2008
Laser desorption/ionization mass spectrometry fordirect profiling and imaging of small moleculesfrom raw biological materialsSangwon ChaIowa State University
Follow this and additional works at: https://lib.dr.iastate.edu/rtd
Part of the Analytical Chemistry Commons, and the Plant Biology Commons
This Dissertation is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State UniversityDigital Repository. It has been accepted for inclusion in Retrospective Theses and Dissertations by an authorized administrator of Iowa State UniversityDigital Repository. For more information, please contact [email protected].
Recommended CitationCha, Sangwon, "Laser desorption/ionization mass spectrometry for direct profiling and imaging of small molecules from rawbiological materials" (2008). Retrospective Theses and Dissertations. 15643.https://lib.dr.iastate.edu/rtd/15643
Laser desorption/ionization mass spectrometry for direct profiling and imaging of small molecules from raw biological materials
by
Sangwon Cha
A dissertation submitted to the graduate faculty
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
Major: Analytical Chemistry
Program of Study Committee: Edward S.Yeung, Major Professor
Robert S. Houk Srdija Jeftinija
Victor S.-Y. Lin Basil J. Nikolau Hans Stauffer
Iowa State University
Ames, Iowa
2008
Copyright © Sangwon Cha, 2008. All rights reserved.
3307070
3307070 2008
Copyright 2008 by Sangwon Cha All rights reserved
ii
To My Families and Friends
iii
TABLE OF CONTENTS
ABSTRACT v
CHAPTER 1. GENERAL INTRODUCTION 1
Dissertation Organization 1 Laser Desorption Ionization Mass Spectrometry 1 Laser Desorption Ionization Mass Spectrometry of Small Molecules 3 Imaging Mass Spectrometry of Small Molecules 4 References 8
CHAPTER 2. IMAGING OF CEREBROSIDES DIRECTLY FROM RAT BRAIN TISSUE BY COLLOIDAL GRAPHITE-ASSISTED LASER DESORPTION/IONIZATION MASS SPECTROMETRY 12
Abstract 12 Introduction 13 Experimental Section 18 Results and Discussion 21 Conclusions 29 Acknowledgements 30 References 34 Figure Captions 40
CHAPTER 3. DIRECT PROFILING AND IMAGING OF PLANT METABOLITES IN INTACT TISSUES BY USING COLLOIDAL GRAPHITE-ASSISTED LASER DESORPTION IONIZATION MASS SPECTROMETRY 53
Abstract 53 Introduction 54 Experimental Section 57 Results and Discussion 61 Acknowledgement 67 References 70 Figure Captions 75
CHAPTER 4. DIRECT PROFILING AND IMAGING OF EPICUTICULAR WAXES ON ARABIDOPSIS THALIANA BY LASER DESORPTION/IONIZATION MASS SPECTROMETRY USING SILVER COLLOID AS A MATRIX 88
Abstract 88 Introduction 88 Experimental Section 90 Results and Discussion 93 Conclusions 102
iv
Acknowledgement 103 References 109 Figure Captions 111
CHAPTER 5. COLLOIDAL SILVER LASER DESORPTION/IONIZATION MASS SPETROMETRY OF STEROLS AND DIRECT PROBING OF CHOLESTEROL ON ASTROCYTE CELL MONOLAYER 124
Abstract 124 Introduction 125 Experimental Section 127 Results and Discussion 130 Conclusions 133 Ackowledgement 133 References 134 Figure Captions 137
CHAPTER 6. GENERAL CONCLUSIONS 144
APPENDIX 1. SUPPORTING FIGURES FOR CHAPTER 2 146
Figure Captions 146
APPENDIX 2. DIRECT PROFILING AND MS IMAGING OF SMALL METABOLITES FROM FRUITES BY COLLOIDAL GRAPHITE-ASSISTED LASER DESORPTION/IONIZATION MASS SPECTROMETRY 155
Abstract 155 Introduction 156 Experimental Section 159 Results and Discussion 161 Acknowledgements 167 References 169 Figure Captions 174
APPENDIX 3. SUPPORTING INFORMATION FOR CHAPTER 3 188
Figure Captions 191
ACKNOWLEDGEMENT 194
v
ABSTRACT
Matrix-assisted laser desorption/ionization(MALDI) mass spectrometry(MS) has
been widely used for analysis of biological molecules, especially macromolecules such as
proteins. However, MALDI MS has a problem in small molecule (less than 1 kDa) analysis
because of the signal saturation by organic matrixes in the low mass region. In imaging MS
(IMS), inhomogeneous surface formation due to the co-crystallization process by organic
MALDI matrixes limits the spatial resolution of the mass spectral image. Therefore, to make
laser desorption/ionization (LDI) MS more suitable for mass spectral profiling and imaging
of small molecules directly from raw biological tissues, LDI MS protocols with various
alternative assisting materials were developed and applied to many biological systems of
interest.
Colloidal graphite was used as a matrix for IMS of small molecules for the first time
and methodologies for analyses of small metabolites in rat brain tissues, fruits, and plant
tissues were developed. With rat brain tissues, the signal enhancement for cerebroside
species by colloidal graphite was observed and images of cerebrosides were successfully
generated by IMS. In addition, separation of isobaric lipid ions was performed by imaging
tandem MS. Directly from Arabidopsis flowers, flavonoids were successfully profiled and
heterogeneous distribution of flavonoids in petals was observed for the first time by graphite-
assisted LDI(GALDI) IMS.
Aqueous-based colloidal silver solution was also investigated as an alternative matrix
for IMS application. For Arabidopsis thaliana wild-type and its cer2 mutant flowers, direct
profiling and imaging of cuticular wax compounds by silver LDI MS and gas
chromatography (GC) - MS were carried out for the first time and cuticular wax metabolites
of them were compared for predicting the function of cer2 gene products. Results from silver
LDI MS showed a good agreement with those from traditional GC-MS analysis and we
vi
propose that cer2 gene mutation mainly affects the conversion from C30 fatty acid to C29
alkane in wax biosynthesis pathway. We also applied silver LDI MS to probe cholesterol,
which is the major component of lipid raft domains, from the Astrocyte cell monolayer. With
colloidal silver, cholesterol was readily detected with high sensitivity and cholesterol level
changes on cell monolayer by methyl β-cyclodextrin treatment were successfully assayed by
introducing relative intensity profiling. Results from more rapid, and simpler silver-LDI MS
cholesterol assay method also showed a good agreement with those from traditional
enzymatic fluorometry cholesterol assay method.
1
CHAPTER 1. GENERAL INTRODUCTION
Dissertation Organization
This dissertation begins with a general introduction of small molecule analysis by
laser desorption/ionization (LDI) mass spectrometry (MS) and imaging mass spectrometry
(IMS). The following four chapters are organized with published papers and a manuscript to
be submitted. Chapters two and three are comprehensive studies of LDI MS and IMS of
small metabolites in intact animal and plant tissues by using colloidal graphite as a matrix.
Chapters four and five are projects related to LDI MS and IMS of small metabolites in plant
tissues and cell monolayer by using colloidal silver as a matrix. Tables, cited literature, and
figures for each chapter were placed at the end of each chapter. Chapter six includes
summary of work and the general conclusions. The following three appendixes include the
extended work related to chapter two and supporting information for chapter two and three.
Laser Desorption Ionization Mass Spectrometry
Since the concept of the mass spectrometry was introduced by J.J Thomson’s cathode
ray tube experiment, mass spectrometry has been developed over 100 years by innovative
researchers and has become one of the most important analysis techniques in almost every
field of natural sciences. Mass spectrometry is the analytical technique measuring mass and
charge state of the analyte. In mass spectrometry (MS), the most important step is the
ionization step because this step makes neutral molecules detectable by converting them to
charged ions. Ionization can be done through various mechanisms such as protonation,
deprotonation, cationization, electron ejection, and electron capture. To achieve ionization
through one or more mechanisms above, many ionization techniques have been developed.
Electron ionization (EI) is the classical ionization technique for MS had and has been
used as the primary ionization techniques in the early stage of MS. EI is based on the
2
interaction between emitted electrons and neutral analytes. In 1960 and 1970’s various
ionization methods such as chemical ionization1, electrospray ionization2, field desorption
ionization3, plasma desorption ionization4 and atmospheric pressure chemical ionization5
techniques were developed for MS. In early 1980’s, inductively coupled plasma was
introduced as an ion source for MS6 and fast atom bombardment technique has also
developed for an ion source.7
Laser sources were demonstrated as a fast surface heating source for generating atom
ions from solid samples8 a few years after discovery of lasing in 1960. Since then, laser-
induced desorption/ionization techniques have been thoroughly investigated as ion sources
for MS. In 1978, polar, nonvolatile organic molecules were ionized and detected by using
submicrosecond pulsed laser.9 Those molecules were ionized through cationization and
detected as sodium or potassium adduct ions.9 In the 1980’s, ultra fine metal powder
(UFMP) which had been used for alloy production was introduced as an assisting material for
laser desorption.10 Compared to bulk material, UFMP has much smaller dimensions which
are in the range of the long light wavelength. Therefore, UFMP could absorb energy more
efficiently from the light source without loss or dispersion of heat and high temperature
required for intact molecule desorption could be achieved by using UFMP as an assisting
material.10 With UFMP, intact polyethylene glycols were detected with very little
fragmentations.11 The use of UFMP for LDI MS was further developed by Tanaka. He used a
UFMP suspension in glycerin as an assisting material for LDI MS, and macromolecules over
100 kDa were observed without fragmentation by this method.12 Inspired by UFMP
suspension matrix for LDI MS, Karas and Hillenkamp further optimized this methodology
for macromolecule analysis13, 14. By using nicotinic acid as a matrix, they observed several
proteins including lysozyme, trypsin, and albumin in the form of molecular ion (MH+).13 LDI
MS with organic acids as matrixes is referred to as matrix-assisted laser desorption/ionization
mass spectrometry, MALDI MS. With elctrospray ionization (ESI) technique15, MALDI MS
3
has become the most widely-used soft-ionization technique for macromolecule analysis,
especially proteins.
Laser Desorption Ionization Mass Spectrometry of Small Molecules
Application of MALDI to small molecule analysis (<1000 Da) seems to be
problematic because of a high degree of signal suppression in the low mass region by
dominant matrix ion signals. However, MALDI MS is still attractive in small molecule
analysis because of the direct sampling capability from complex mixtures and the tolerance
for contaminants. With successes in interfacing between MALDI sampling part and a variety
of mass analyzers, the instrumental capability of LDI mass spectrometers such as mass
spectral resolution and tandem MS capability is no longer a limiting factor any more in small
molecule analysis. In addition, great achievements in alternative sample preparation methods
for LDI MS allow us to detect small molecules with less severe matrix interference and/or
sample inhomogeneity.
First, organic matrixes which have relatively large molecular weights and therefore
there was no interference present in the low mass region were used. For example, meso-
tetrakis(pentafluorophenyl) porhyrin (MW. 974.46) was used as a matrix to analyze low
molecular weight alkylphenol ethoxylates which range from m/z 300 to m/z 800.16 Second,
alternative inorganic matrixes such as cobalt powder12, silver colloids17, gold nanoparticles18
and carbon particles were used to analyze small molecules. Among those, carbon particles
including graphite, carbon powder, and carbon nanotubes showed relatively clean
background which is suitable for small molecule analysis. Examples of small molecule
analysis by using carbon materials are reviewed briefly in this dissertation at the introduction
part in Chapter 2. Recently, 30nm-sized silicon nanoparticles were used as a matrix to
analyze various drug, small peptide, and pesticide molecules.19 Interestingly, much lower
laser fluence was required for the silicon nanoparticle matrix compared to conventional
4
MALDI matrixes.19 Third, electrochemically etched porous silicon surfaces without any
additional matrix were used for small molecule analysis by LDI MS. This technique is
referred to as desorption/ionization on silicon MS (DIOS MS).20 Various classes of small
molecules including carbohydrates were analyzed with very low detection limit and with low
background.21-23 Further development on DIOS has been made by utilizing silicon nanowires
as sample support and it required much lesser laser fluence than porous silicon surface or
conventional MALDI environment.24 Lastly, nanostructures coated with liquid-phase
perfluorinated initiator molecules were used as sample support for both LDI MS and SIMS.25
This technique is called as ‘nanostructure initiator MS (NIMS)’. NIMS showed greater
sensitivity (~tens of attomoles) over MALDI and ESI in detecting small molecules including
lipids and drug molecules.25
Imaging Mass Spectrometry of Small Molecules
Pulsed laser-based or energetic particle-based techniques in mass spectrometry have
the capability of generating ions directly from the specific location of sample surfaces. This
capability of spatial sampling allowed us to investigate chemical spatial distribution by
performing serial data collection through sample surfaces. This ‘microscope-concept’
approach by mass spectrometry is called as ‘imaging mass spectrometry (IMS)’ or mass
spectral imaging’. Major methodologies used for IMS are MALDI MS26 and secondary ion
MS (SIMS)27. MALDI IMS was first introduced by Caprioli group.28 MALDI IMS is
advantageous over SIMS in detecting intact large molecules (>1000 Da). However, since the
spatial resolution is limited by laser spot size in MALDI IMS, a spatial resolution of MALDI
(10 to hundreds of microns) is worse than that of SIMS (~100 nm). To get higher spatial
resolution in MALDI IMS, decreasing the laser spot size close to the diffraction limit29 or
using astigmatic optics with a position-sensitive detector30, 31 were demonstrated. In addition,
homogeneous deposition of matrixes is critical in MALDI IMS because all conventional
5
MALDI matrixes crystallize inhomogeneously on sample surfaces. For homogeneous matrix
deposition, finely controlled matrix spraying devices such as acoustic droplet ejection
instrument and piezoelectric microdispenser have been developed.32-34
Because of the soft-ionization characteristic of MALDI, MALDI IMS has been
mainly used for studying protein distribution from tissue sections35-37. However, small
molecules below 1 kDa such as drug molecules38, 39 and endogenous primary or secondary
metabolites were also spatially investigated from many biological systems by MALDI IMS.
Among various primary and secondary metabolites, lipid is the major class of compounds in
small molecule imaging by MALDI IMS because of their abundances in tissue samples.
However, there are two major challenging issues in IMS of small molecules. First, peaks
from conventional MALDI matrixes are dominant in the low mass region and these peaks
interfere with or suppress signals from small sample molecules. To overcome this, matrix-
free methodology such as IR-MALDI was applied to profile plant metabolites40, or matrix-
embedded surfaces such as clathrate nanostructures25 and silicon41 were employed to image
small molecules from intact tissues.
Second, several isobaric ions can be associated in one nominal m/z value which can
cause wrong or overlapped spatial information for a specific compound when generation of
mass spectral images was simply based on extracting intensity values from mass spectra.
Because of compositional similarities of small metabolites, there are many possible overlaps
by several different compounds in the low mass region. To separate distributions of isobaric
ions, imaging tandem MS (IMS/MS)42, high mass resolution MS36, and ion mobility
spectrometry43 were employed.
In IMS/MS, first generation product ions which were specifically from only one of
isobaric ions were chosen for generating chemically selective images. For example, both
sodiated phosphatidylcholine 38:6 ions ([PC 38:6 + Na]+) and DHB matrix cluster ions
contributed the peak intensity at m/z 828.44 But distribution of these two ions were
6
successfully separated by generating images of their specific product ions at m/z 769 (neutral
loss of 59 Da, loss of PC headgroup) and at m/z 652 (loss of a sodiated DHB molecule).44
One major disadvantage of IMS/MS is that it is a time-consuming process because a tissue
sample needs to be scanned separately for all single m/z values which were associated with
several isobaric ions.
High mass spectral resolution and accurate mass measurement by using MALDI
fourier transform ion cyclotron resonance(FTICR) MS can resolve isobaric ions which
cannot be resolved by low-mass resolution TOF or ion-trap mass spectrometers. For example,
Images for three peptides within 0.24 Da range at m/z 1293 were successfully obtained from
the rat brain tissue.36 Mass accuracy was 6 ppm in this experiment.36 However, high mass
resolution imaging is also time-consuming process because of longer measuring time by FT-
ICR.
Ion mobility spectrometry is a real-time gas phase ion separation technique.45, 46
Because ion mobility separation was performed prior to mass analysis, detected ions bear the
information of ions mobility drift time as well as their m/z values. Isobaric ions from
different classes can be separated by choosing different drift time region. For example, lipids
detected from rat brain tissues showed about 12% longer drift time than isobaric peptides.47
Therefore, signals from lipids were separated from chemical noises and other isobaric ion
signals by choosing a specific mobility drift time.43 However, acquiring mass and drift time
are limited in the current software and hardware instrumentation of ion mobility spectrometry
because the allowed data capacity is not enough to process all ions collected from one
scanning point.43, 47
7
(a) H
igh
Mas
s R
esol
utio
n(b
) Ion
Mob
ility
Pro
duct
ion
atm/z
773
(NL
59)
Pro
duct
ion
atm/z
670
(NL
162)
0%10
0%
m/z
832
(c) M
S2Im
agin
g
Figure 1. Isobaric ion separations in IMS. Isobaric ions can be separated by (a) high
mass spectral resolution MS36, (b) ion mobility MS43, (c) tandem MS42.
8
References
(1) Munson, M. S. B.; Field, F. H. Journal of the American Chemical Society 1966, 88,
2621-&.
(2) Dole, M.; Mack, L. L.; Hines, R. L. Journal of Chemical Physics 1968, 49, 2240-&.
(3) Beckey, H. D. Angewandte Chemie-International Edition 1969, 8, 623-&.
(4) Macfarlane, R. D.; Torgerson, D. F. Science 1976, 191, 920-925.
(5) Carroll, D. I.; Dzidic, I.; Stillwell, R. N.; Haegele, K. D.; Horning, E. C. Analytical
Chemistry 1975, 47, 2369-2373.
(6) Houk, R. S.; Fassel, V. A.; Flesch, G. D.; Svec, H. J.; Gray, A. L.; Taylor, C. E.
Analytical Chemistry 1980, 52, 2283-2289.
(7) Barber, M.; Bordoli, R. S.; Sedgwick, R. D.; Tyler, A. N. Journal of the Chemical
Society-Chemical Communications 1981, 325-327.
(8) Honig, R. E.; Woolston, J. R. Applied Physics Letters 1963, 2, 138-139.
(9) Posthumus, M. A.; Kistemaker, P. G.; Meuzelaar, H. L. C.; Tennoeverdebrauw, M. C.
Analytical Chemistry 1978, 50, 985-991.
(10) Tanaka, K. In Les Prix Nobel, The Nobel Prizes 2002; Frängsmyr, T., Ed.; Nobel
Foundation: Stockholm, 2002, pp 197-217.
(11) Wilkins, C. L.; Weil, D. A.; Yang, C. L. C.; Ijames, C. F. Analytical Chemistry 1985,
57, 520-524.
(12) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T.; Matsuo, T. Rapid
Communications in Mass Spectrometry 1988, 2, 151-153.
(13) Karas, M.; Hillenkamp, F. Analytical Chemistry 1988, 60, 2299-2301.
(14) Karas, M.; Ingendoh, A.; Bahr, U.; Hillenkamp, F. Biomedical and Environmental
Mass Spectrometry 1989, 18, 841-843.
(15) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Science 1989,
246, 64-71.
9
(16) F. O. Ayorinde, P. H. T. N. P. Q. L. K., Jr. Rapid Communications in Mass
Spectrometry 1999, 13, 2474-2479.
(17) Sluszny, C.; Yeung, E. S.; Nikolau, B. J. Journal of the American Society for Mass
Spectrometry 2005, 16, 107-115.
(18) McLean, J. A.; Stumpo, K. A.; Russell, D. H. J. Am. Chem. Soc. 2005, 127, 5304-
5305.
(19) Wen, X.; Dagan, S.; Wysocki, V. H. Anal. Chem. 2007, 79, 434-444.
(20) Wei, J.; Buriak, J. M.; Siuzdak, G. Nature 1999, 399, 243-246.
(21) Trauger, S. A.; Go, E. P.; Shen, Z. X.; Apon, J. V.; Compton, B. J.; Bouvier, E. S. P.;
Finn, M. G.; Siuzdak, G. Analytical Chemistry 2004, 76, 4484-4489.
(22) Compton, B. J.; Siuzdak, G. Spectroscopy-an International Journal 2003, 17, 699-
713.
(23) Want, E.; Compton, B. J.; Hollenbeck, T.; Siuzdak, G. Spectroscopy-an International
Journal 2003, 17, 681-691.
(24) Go, E. P.; Apon, J. V.; Luo, G.; Saghatelian, A.; Daniels, R. H.; Sahi, V.; Dubrow,
R.; Cravatt, B. F.; Vertes, A.; Siuzdak, G. Analytical Chemistry 2005, 77, 1641-1646.
(25) Northen, T. R.; Yanes, O.; Northen, M. T.; Marrinucci, D.; Uritboonthai, W.; Apon,
J.; Golledge, S. L.; Nordstrom, A.; Siuzdak, G. Nature 2007, 449, 1033-1036.
(26) Cornett, D. S.; Reyzer, M. L.; Chaurand, P.; Caprioli, R. M. Nature Methods 2007, 4,
828-833.
(27) Heeren, R. M. A.; McDonnell, L. A.; Amstalden, E.; Luxembourg, S. L.; Altelaar, A.
F. M.; Piersma, S. R. Applied Surface Science 2006, 252, 6827-6835.
(28) Caprioli, R. M.; Farmer, T. B.; Gile, J. Analytical Chemistry 1997, 69, 4751-4760.
(29) Spengler, B.; Hubert, M. Journal of the American Society for Mass Spectrometry
2002, 13, 735-748.
10
(30) Luxembourg, S. L.; McDonnell, L. A.; Mize, T. H.; Heeren, R. M. A. Journal of
Proteome Research 2005, 4, 671-673.
(31) Luxembourg, S. L.; Mize, T. H.; McDonnell, L. A.; Heeren, R. M. A. Analytical
Chemistry 2004, 76, 5339-5344.
(32) Aerni, H. R.; Cornett, D. S.; Caprioli, R. M. Analytical Chemistry 2006, 78, 827-834.
(33) Bogan, M. J.; Agnes, G. R. Rapid Communications in Mass Spectrometry 2004, 18,
2673-2681.
(34) Stoeckli, M.; Staab, D.; Staufenbiel, M.; Wiederhold, K. H.; Signor, L. Analytical
Biochemistry 2002, 311, 33-39.
(35) DeKeyser, S. S.; Kutz-Naber, K. K.; Schmidt, J. J.; Barrett-Wilt, G. A.; Li, L. J.
Journal of Proteome Research 2007, 6, 1782-1791.
(36) Taban, I. M.; Altelaar, A. F. M.; Van der Burgt, Y. E. M.; McDonnell, L. A.; Heeren,
R. M. A.; Fuchser, J.; Baykut, G. Journal of the American Society for Mass
Spectrometry 2007, 18, 145-151.
(37) Verhaert, P. D.; Conaway, M. C. P.; Pekar, T. M.; Miller, K. International Journal of
Mass Spectrometry 2007, 260, 177-184.
(38) Khatib-Shahidi, S.; Andersson, M.; Herman, J. L.; Gillespie, T. A.; Caprioli, R. M.
Analytical Chemistry 2006, 78, 6448-6456.
(39) Hsieh, Y.; Casale, R.; Fukuda, E.; Chen, J. W.; Knemeyer, I.; Wingate, J.; Morrison,
R.; Korfmacher, W. Rapid Communications in Mass Spectrometry 2006, 20, 965-972.
(40) Li, Y.; Shrestha, B.; Vertes, A. Analytical Chemistry 2008, 80, 407-420.
(41) Liu, Q.; Guo, Z.; He, L. Analytical Chemistry 2007, 79, 3535-3541.
(42) Cha, S. W.; Yeung, E. S. Analytical Chemistry 2007, 79, 2373-2385.
(43) McLean, J. A.; Ridenour, W. B.; Caprioli, R. M. Journal of Mass Spectrometry 2007,
42, 1099-1105.
11
(44) Garrett, T. J.; Prieto-Conaway, M. C.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford,
G.; Yost, R. A. International Journal of Mass Spectrometry 2007, 260, 166-176.
(45) McLean, J. A.; Ruotolo, B. T.; Gillig, K. J.; Russell, D. H. International Journal of
Mass Spectrometry 2005, 240, 301-315.
(46) Vonhelden, G.; Wyttenbach, T.; Bowers, M. T. Science 1995, 267, 1483-1485.
(47) Jackson, S. N.; Ugarov, M.; Egan, T.; Post, J. D.; Langlais, D.; Schultz, J. A.; Woods,
A. S. Journal of Mass Spectrometry 2007, 42, 1093-1098.
12
CHAPTER 2. IMAGING OF CEREBROSIDES DIRECTLY FROM RAT
BRAIN TISSUE BY COLLOIDAL GRAPHITE-ASSISTED LASER
DESORPTION/IONIZATION MASS SPECTROMETRY
A paper published in Analytical Chemistry*
Sangwon Cha and Edward S. Yeung
Abstract
Graphite assisted laser desorption/ionization (GALDI) mass spectrometry (MS) was
investigated for analysis of cerebrosides in complex total brain lipids extract. Conventional
MALDI MS and graphite GALDI MS were compared regarding lipid analysis by using high-
vacuum (HV, <10-6 Torr) LDI time-of-flight (TOF) mass spectrometry and intermediate
pressure (IP, 0.17 Torr) linear ion trap (LIT) mass spectrometry. Cerebrosides were not
detected or detected with low sensitivity in MALDI MS because of other dominant
phospholipids. By using GALDI, cerebrosides were detected as intense mass peaks without
prior separation from other lipid species while mass peaks corresponding to
phosphatidylcholines (PCs) were weak. The signal increase for cerebrosides and the signal
decrease for PCs in GALDI MS were more significant in HV than in IP. MSn experiments of
precursor ions corresponding to cerebrosides and PCs in brain lipid extract were performed
for identifying the detected species and for distinguishing isobaric ions. 22 cerebroside
___________________________________________________________________________
* Reprint with permission from Analytical Chemistry 2007, 79(6), 2373-2385.
Copyright © 2007 American Chemical Society
13
species were detected by GALDI whereas 8 cerebroside species were detected by MALDI.
Sulfatides in brain lipid extract were also easily detected by GALDI MS in the negative ion
mode. By forming a colloidal graphite thin film on rat brain tissue, direct lipid profiling by
imaging mass spectrometry (IMS) was performed. Chemically-selective images for
cerebrosides and sulfatides were successfully obtained. Imaging tandem mass spectrometry
(IMS/MS) was performed to generate images of specific product ions from isobaric species.
Introduction
Cerebrosides are one class of glycosphingolipids. A cerebroside is composed of a
hexose (galactose or glucose) and a ceramide consisting of a sphingosine or a sphinganine
base and an amide-linked long chain fatty acid. Depending on the hexose unit, a cerebroside
can be called a galactosylceramide (GalCer) or a glucosylceramide (GlcCer). As mediators,
cerebrosides are involved in many biological processes, such as cell adhesion, cell growth,
cell morphogenesis, and cell-to-cell communication.1-3 Cerebrosides also act as receptors or
as binding sites for viruses on a cellular membrane.4 Deficiency of GalCer causes abnormal
function and local instability during myelination.5 Abnormal accumulation of GalCer or
GlcCer also results in Krabbe disease or Gaucher’s disease, respectively. This increased level
of GalCer or GlcCer is caused by a deficiency of galactosylceramidase or
glucocerebrosidase.6 Krabbe disease causes severe mental degeneration because of
insufficient development of myelin and Gaucher’s disease leads to bone damage, anemia, and
enlargement of spleen and liver.6 In brain tissue, GalCer species are abundant,7, 8 at up to 40
mol% of the phosphatidylcholines (PCs) in rat brain cortex depending on their locations.8-10
Among cortex parts, GalCers are 8 to 10 times more enriched in white matter than in gray
matter.3 In addition to GalCers, 2-hydroxy GalCer species are also abundant in mammalian
brain tissues and their total amount is about 50% of GalCer species.3
14
Since soft ionization techniques for mass spectrometry (MS) such as matrix-assisted
laser desorption/ionization (MALDI) and electrospray ionization (ESI) have been
introduced,11-13 many researchers applied these techniques to analyze lipids. Excellent recent
reviews about lipid analysis by using MALDI or ESI technique can be found elsewhere.10, 14,
15 Lipid analysis with MALDI has the potential advantage over ESI in sample preparation
because of the possibility of using raw biological materials such as tissue sections directly
(without extraction) to obtain spatial information. MALDI MS, however, has a problem with
analyzing lipid mixtures because ionization efficiencies are different according to their head
groups. The presence of PCs and sphingomyelines (SMs) which have a quaternary
ammonium polar head group has been known to suppress signals of other lipid species in the
positive ion mode.16 For lipid analysis with MALDI MS, the most common organic matrix is
2,5-dihydroxy benzoic acid (DHB).17-19 Using DHB with the addition of cesium ion (Cs+)
resulted in reducing spectra complexity and enhancing sensitivity.20, 21 Other matrixes, such
as 2,6-dihydroxyacetophenone (DHA), sinapinic acid (SA), α-cyano-4-hydroxycinnamic
acid (CHCA), p-nitroaniline (PNA), and others, have also been tested,17, 19-21 PNA
generated good profiles of PEs in the presence of PCs and SMs.20 CHCA and SA, however,
suffered from interference signals originated from matrixes. Their low solubility in methanol
was also problematic.14, 17
Whereas numerous MALDI MS analyses of phospholipids have been reported and
many MALDI MS analyses of gangliosides which consists of a glycosphingolipid and one or
more sialic acids have been performed,19, 22-28 relatively little research has been carried out
on the analyses of cerebrosides by using MALDI MS despite their biological importance.
Harvey examined comprehensively the applicability of MALDI MS to sphingolipids and
glycosphingolipids.18 Analysis of less-polar neutral and acidic cerebrosides’ derivatives in
monkey brain was performed by MALDI MS after serial chromatographic separations of
cerebroside-related compounds.29 Yurkova and coworkers induced free-radical
15
fragmentations of cerebrosides by irradiation of γ-ray of cerebroside-containing micelles and
analyzed the lipid extracts from these irradiated micelles by MALDI MS.30 Fujiwaki and
coworkers analyzed sphingolipids in tissues and body fluids from patients with Gaucher’s
disease and found that cerebrosides/SM ratio was elevated in samples from patients with
Gaucher’s disease.23, 31 However, all analyses of cerebrosides mentioned above required one
or more separation steps for isolating glycosphigolipids from other lipid species before MS
analysis.23, 29, 31 Without separation, it is hard to detect cerebrosides by conventional
MALDI MS because phospholipids, especially PCs, are usually dominant in crude extracts
from eukaryotes and the m/z region of phospholipids in the mass spectra substantially
overlaps that of cerebrosides.
Many types of graphite matrices have been suggested as alternative matrices for laser
desorption/ionization (LDI) mass spectrometry because they are near blackbody and disperse
the energy very effectively. Graphite has been used in forms of suspension in glycerol32, 33
or 2-propanol,34 powder35 and plate.36-39 These particles or flakes can range from 2 µm to
150 µm. Sunner and coworkers first showed that graphite matrix in glycerol helped the
detection of proteolytic digest of cytochrome c and cytochrome c itself.32 Further
development of graphite/liquid matrix was achieved by Dale and coworkers and proteins,
oligosaccharides, and synthetic polymers were successfully detected with a glycerol/graphite
matrix.33 The aging of the triterpene in mastic and dammar films was also studied with
graphite powder as a matrix.35 With graphite plate, poly(methylsilsesquioxane)s,38 fatty
acids,39 and synthetic polymers36, 37 were analyzed. Peng and coworkers developed an
interface for thin-layer chromatography (TLC) plate scanning by atmospheric pressure (AP)
LDI MS. They used graphite particle suspension in 2-propanol as a matrix for detecting SMs
and PCs directly from the TLC plate.34 Graphite trapped in silicon polymer was also used for
the surface support to facilitate detecting peptides and proteins with desalting effect.40 Pencil
leads, which are mixed forms of graphite, wax, and clay were used as a matrix to analyze
16
actinide materials and as a calibrant.41 Activated carbon powder was used for detecting
various organic compounds including peptides, prometryn, and hydrochlorothiazide on the
TLC plate.42 Functionalized fullerenes, C60((CH2)2COOH)n or C60(C11H23)n, were also
used for the detection of peptides and phospholipids.43 Recently, carbon nanotubes or
oxidized carbon nanotubes were suggested as matrixes for detecting small peptides, small
oligosaccharides, and cyclodextrins.44, 45 Oxidized carbon nanotubes, which have carboxyl
groups on oxidized surfaces of carbon nanotubes, have more polar surfaces and increased
solubility in aqueous solution.46, 47
For imaging mass spectrometry (IMS), several ionization techniques such as matrix
enhanced secondary ion mass spectrometry (ME-SIMS)48, metal-assisted SIMS,49, 50
MALDI, and cluster SIMS51 have been used. The advantages and disadvantages of these
ionization methods for IMS are well described in the recent review article.52 Briefly, SIMS
has the advantage in spatial resolution but MALDI is more suitable for detecting intact
biological macromolecules. MALDI is also advantageous in terms of instrumental simplicity
and cost. IMS with MALDI has been applied to biomarker studies and many other
pathologically interesting case studies. 53-58
There are two major issues in MALDI IMS. First, spatial resolution is usually limited
by the laser spot size, which usually ranges from 80 µm to 200 µm in diameter. Spengler and
coworkers decreased the spot size to about 0.6 µm by modifying the focusing lens.59
Similarly, Luxembourg and coworkers utilized astigmatic optics and a position-sensitive
detector.60, 61 However, the spatial resolution of 4 µm from this system was determined by
the 2D-position sensitive detector and not by the laser spot size.
Second, organic matrices used in conventional MALDI need co-crystallization that
can lead to surface inhomogeneity. Luxembourg and coworkers showed different spectral
profiles throughout the DHB-spotted surface by ME-SIMS.62 Spraying a matrix solution
electrically or pneumatically, immersing the sample in matrix solution, and dispensing matrix
17
solution by a piezoelectric microdispenser were suggested for generating homogeneous
sample surface with matrixes.63-65 Recently, Aerni and coworkers used acoustic droplet
ejection66 to generate 180-200 µm matrix droplets reproducibly on tissues. The other
solution for generating matrix-sample surface is using ionic matrices. Armstrong and
coworkers introduced new ionic liquids, which are basically combinations of acidic MALDI
matrixes (e.g., DHB, SA, CHCA and etc.) and basic organic solvents such as pyridine, as
MALDI matrixes.67 Ionic liquid matrixes showed advantages of homogeneity, signal
reproducibility, and vacuum-stability.68-70 Lemaire and coworkers improved the
performance of MALDI MS and IMS on rat brain tissue by using solid ionic matrixes such as
CHCA/aniline in the negative-ion mode.71 Recently, imaging tandem mass spectrometry
(IMS/MS) of rat brain tissues by using IP-MALDI LIT mass spectrometer has been
reported.72 By doing MS/MS for site-specifically generated ions, separate chemically-
selective images of product ions were generated for the isobaric ions at m/z 828.6 which
could be composed of sodiated PC 38:6, DHB cluster, and phosphatidylethanolamine 38:4
(PE 38:4).72
Here we demonstrated a simple approach for GALDI MS by using an aerosol spray
which contains 2-propanol-based colloidal graphite. Colloidal GALDI and conventional
MALDI mass spectral analyses were performed for standard lipid mixtures which contain
various compositions of PCs and GlcCers in high-vacuum (HV, 10-6 Torr) and at
intermediate pressure (IP, 0.17 Torr). Softer ionization induced by collisional or vibrational
cooling in IP-MALDI resulted in less fragmentation for labile molecules including
gangliosides.24, 25 By performing imaging MS of sample spots prepared by DHB matrix, we
can derive location-specific mass spectral profiles for PC/GlcCers lipid mixtures as in earlier
studies.62, 73, 74 Total brain lipid extract was chosen as a model of a complex lipid mixture
and was analyzed under four different sampling conditions. In addition, GALDI MSn
experiments were performed to identify the ionized species and to examine isobaric ions. We
18
also investigated the applicability of the colloidal graphite spraying method to direct rat brain
tissue analyses and IMS. Finally, overlapping spatial information originating from PC and
GalCer isobaric ions were separated from each other by performing IMS/MS.
Experimental Section
1,2-Dipalmitoyl-sn-glycero-3-phosphorylcholine (PC 32:0) and standard GlcCers
from bovine buttermilk were purchased from Matreya LLC (Pleasant Gap, PA). Fatty acid
composition of GlcCers from bovine buttermilk provided by the manufacturer is 14:0 (trace),
16:0 (15%), 18:0 (3%), 20:0 (2%), 22:0 (31%), 23:0 (28%), 24:0 (17%), and others (4%).
Long chain bases for these GlcCers are all C18-sphingosine (d18:1) bases. Brain total lipid
extract in chloroform solution (10 mg/ml) was obtained from Avanti Polar Lipids Inc.
(Alabaster, AL). DHB from Bruker Daltonics (Billerica, MA) was used. 2-propanol-based
colloidal graphite aerosol spray (Aerodag® G) was obtained from Acheson Colloids (Port
Huron, MI). All other chemicals were obtained from Fisher Scientific (Fairlawn, NJ).
Rat brain was collected from Spargue-Dawley rats from the College of Veterinary
Medicine at Iowa State University and was frozen in liquid N2 immediately. The frozen rat
brain was cryosectioned into 15 µm thick specimens without embedding in the optimum
cutting temperature (OCT) compound. Without the OCT compound, mass spectral
interference is known to be decreased.63 A cryostat from International Equipment Co.
(Needham Heights, MA) was used for cryosectioning. Sectioned tissues were directly
transferred and mounted onto a stainless-steel plate. The prepared tissue sections were stored
at –20 °C before mass spectrometric analysis.
HV (<10-6 Torr) LDI mass-spectrometric analysis was done with the Biosystems
Voyager-DE PRO time-of-flight (TOF) mass spectrometer (Framingham, MA) equipped
with a N2 laser (337 nm, and maximum 180 µJ/pulse). Mass spectra were collected both in
positive ion mode and negative ion mode with a ±20 kV accelerating potential. Mass
19
calibration was performed with lipid standards. An LTQ linear ion trap mass spectrometer
equipped with vMALDI source (Thermo Electron, Mountain View, CA) was used for IP
(0.17 Torr) LDI mass-spectrometric analysis and imaging MS. A description of the LTQ with
vMALDI source can be found elsewhere.75, 76 The laser source in this mass spectrometer is a
fiber-optic guided N2 laser (337 nm, and maximum 280 µJ/pulse before entering the optical
fiber cable). The measured laser spot size is about 100 µm in diameter on the sample plate
surface.
For lipid standard mixtures, 1 mg/ml of PC 32:0 and 1 mg/ml of total GlcCers stock
solution were prepared in chloroform and in 70% chloroform/ 30% methanol respectively.
Two types of lipid standard mixture batches were prepared. One batch contained a constant
concentration of PC 32:0 (0.5 mg/ml) and various concentrations of GlcCers (from 0.005
mg/ml to 0.5 mg/ml). In the other batch, the concentration of GlcCers was kept constant (0.5
mg/ml), but the concentration of PC 32:0 were from 0.005 mg/ml to 0.5 mg/ml. The total
lipid concentrations in the mixtures thus varied from 0.505 mg/ml to 1 mg/ml. Total brain
lipid extract was diluted with chloroform and the final total lipid concentration was 2 mg/ml.
50 mg/ml DHB solution in 70% methanol and 30% water (containing a 0.1% trifluoroacetic
acid) was prepared as a conventional MALDI matrix.
For conventional MALDI MS, 1 µl of sample solution was applied to the stainless
sample plate followed by 1 µl of DHB matrix solution. For GALDI MS of PC/GlcCer
mixtures and brain lipid extracts, colloidal graphite was sprayed directly onto the stainless-
steel sample plate by using the aerosol spray can (Aerodag® G). The distance between the
nozzle of the spray can and the sample plate was kept at 25 cm. Spraying was performed for
10 to 15 s to form a thin graphite film on the sample plate surface. After spraying, the surface
was dried in air at room temperature for 5 min before the application of sample solutions. For
the standard lipid mixture or total brain lipids extract, 1 µl of sample solution was applied
onto the graphite-coated surface by using a micropipette. For brain tissues, colloidal graphite
20
was diluted four times with 2-propanol and the diluted solution was loaded into a sample
bottle for an airbrush. Double-action airbrush, model “Aztek A470” with a 0.30 mm nozzle
from Testor (Rockford, IL), was used. Spraying with 20 psi air pressure was performed 15.4
cm away from the sample plate. This way, the whole rat brain tissue was covered with
colloidal graphite without moving either the airbrush or the sample plate. Spraying for 30 to
40 s gave the strongest signals from analytes and the lowest graphite background signals.
After spraying, samples were allowed to dry in air for 10 min.
In HV-MALDI MS, mass spectra were recorded from the “big, needle-like crystals”
area at the boundary of the sample spot and from the “small crystals” area at the center of the
sample spot separately. These two mass spectra were then averaged and the averaged mass
spectrum was used for data interpretation. In the case of IP-MALDI MS, rastering over a
sample spot was performed. First, for one sample spot, serial optical images were taken every
1 mm-movement of the sample stage in either x- or y-directions. Each segment of the images
has a size of 140 pixels by 170 pixels. These segments of optical images were reconstructed
as one optical image for one sample spot. Second, a sample spot were rastered with a 100 µm
step size. For each spot, a mass spectrum was recorded for desorbed ions by co-adding over
five laser shots. For each MALDI sample spot, a mass spectrum constructed by averaging
mass spectra from all rastering points was used for data interpretation.
For IMS of rat brain tissue, the number of laser shots for each sample spot was first
determined by collecting mass spectra in a small area (usually 10 rastering points) with
automatic gain control (AGC, which keeps ion amounts in the trap constant by changing the
number of laser shots). Based on the average number of shots per spot with AGC, the number
of laser shots was fixed without AGC (usually 5 to 8 shots per spot) when collecting mass
spectra over the whole tissue. The tissue sample was rastered with 100 µm steps. In the case
of IMS/MS, target precursor ions were first selected based on first generation product ion
spectra for the major peaks in the extracted mass spectra and then first generation product ion
21
spectra of the selected precursor ion were collected with spatial information for all rastering
points over the tissue section (instead of collecting full mass profiles). For a 1.8 cm × 0.8 cm
tissue sample, about 10,000 mass spectra were collected over 1.7 h for IMS and 3.1 h for
IMS/MS, respectively.
Chemically-selective images and extracted mass spectra from specific locations were
generated by using the custom software from Thermo (vMALDI Data Browser). Chemically-
selective images were plotted as maps using either absolute intensity values or normalized
intensity values. The normalized intensity is defined as the fractional peak intensity for the
peak of interest compared to the total ion current (TIC) of each mass spectrum. The mass
window for generating chemically-selective images was 0.8 Da.
Results and Discussion
Colloidal graphite surface for GALDI MS. The electron microscopic image of the
graphite surface generated by colloidal graphite aerosol spray is shown in Figure 1. Graphite
particles in the aerosol spray are about 1 µm-sized with flake shapes (Figure 1) and are finer
than most other types of preparations. The graphite surface generated in this manner has
advantages over other methods of generating graphite surfaces. First, this method is fast and
has minimum preparation steps. Only cleaning of the substrate surface (a stainless steel
sample plate), spraying colloidal graphite, and drying the plate in the air are needed for MS.
It took less than 10 min for the whole preparation. Second, after performing MS, the graphite
surface was easily removed by washing with 2-propanol or acetone. The cleaned plate can be
reused for either GALDI MS or conventional MALDI MS. Third, because of using
conventional stainless steel sample plate as a base, the mechanical stability of the sample
plate, which could be an issue when using a pressurized graphite plate, is very good. Last,
because the colloidal graphite spray contains thermoplastic resin as a binder, formation of
graphite dust is much less problematic than when using a graphite/liquid suspension.
22
MALDI and GALDI MS of mixtures of PC and GlcCers at different sample
chamber pressures. As mentioned before, the presence of PCs and SMs prevents the
detection of other lipid species in MALDI MS. To examine whether this phenomenon occurs
in GALDI MS, standard batch solutions of various compositions of PC 32:0 and GlcCers
were analyzed by MALDI MS and GALDI MS at two different sample chamber pressures.
The mass spectra are shown in Figure 2 and mass peak assignments are listed in Table 1 with
relative abundances. With DHB matrix, protonated PC ions are dominant at both pressures
(Figure 2(a), 2(b), and Table 1). However, with colloidal graphite, the sodiated GlcCer ions
were the most intense at both pressures (Figure 2(c), 2(d), and Table 1). The signal intensity
ratios among different lipid species were different depending on sample chamber pressures.
PC to GlcCer signal intensity ratios were higher in IP than in HV with DHB matrix (Figure
2(a), 2(b), and Table 1). This tendency was even more significant with colloidal graphite
(Figure 2(c), 2(d), and Table 1). In other words, the decrease of PC-related mass peaks or the
increase of GlcCer-related mass peaks was more distinct in HV than in IP.
To examine the ion signals as a function of lipid concentration and composition,
various ratios of PC/GlcCers mixtures were analyzed under the four different conditions.
Relative intensities for the major peak of each lipid species are listed in Table 2, and mass
spectra for these lipid standard mixtures are shown in Supporting Information Figure S-1, S-2,
S-3, and S-4. First, the mass peak corresponding to [PC 32:0 + H]+ was dominant in both
HV- and IP-MALDI mass spectra even when the concentration of PC 32:0 was tenths of that
of total GlcCers (Batch 1). Second, in MALDI, no obvious difference between in IP and in
HV was observed in terms of the relative peak intensities. Third, in contrast to HV- and IP-
MALDI, significant peak intensity for the mass peak corresponding to [GlcCer 22:0 + Na]+
was observed in both HV- and IP-GALDI (Batch 2). Fourth, in GALDI, the peak suppression
of PC 32:0 related ions was less in IP than HV (Batch 1) and the peak enhancement for
[GlcCer 22:0 +Na]+ was higher in HV than IP (Batch 2). In other words, IP-GALDI MS may
23
be more suitable for the simultaneous analysis of various lipid species in complex mixtures,
while HV-GALDI MS may be better for the analysis of cerebrosides in complex lipid
mixtures without separation.
Localization of analytes with DHB matrix. Mass spectral studies for elucidating the
incorporation of analytes into DHB matrix crystals and the heterogeneity of MALDI samples
prepared with DHB matrix have already been reported by many researchers.62, 73, 74 Because
PCs and GlcCers showed very different characteristics in ionization, localization of analytes
was examined by LDI imaging MS. Because the size of conventional laser beams at the
target plate (80-200 µm) is bigger than those of most crystals, resolving one matrix crystal
from another is hard to achieve. However, due to the deposition/drying process, there are two
distinct regions which can be macroscopically recognized from the optical image shown as
the first image in Figure 3(a). One region is located at the boundary of the sample spot and is
mainly composed of “big, needle-like crystals”. The other region is at the center of the
sample spot and “small crystals” are distributed heterogeneously throughout this region. The
subsequent images in Figure 3(a) represent chemically-selective images for the major ions
detected from the PC 32:0/GlcCers mixture. As Luxembourg and coworkers pointed out,62,
73, 74 protonated molecules, [PC 32:0 + H]+, showed the most intense peaks at the boundary
of the sample which consists of big matrix crystals. However, sodiated molecules, such as
[PC 32:0 + Na]+, [GlcCer 22:0 + Na]+, and [GlcCer 23:0 + Na]+ showed much higher peak
intensities at the interior of the MALDI sample spot. This phenomenon also can be seen
clearly from the mass spectra extracted from the two different regions (Figure 3(b) and 3(c)).
This indicates that incorrect mass profiles of lipid mixtures could be generated when spectra
was collected and averaged for a certain region but not for the whole sample spot prepared
with DHB matrix. In contrast to DHB matrix, colloidal graphite does not involve any
crystallization process and are much finer in dimension. Therefore, the surface formed by
24
colloidal graphite provides uniform signals without hot spots, and is well suited for chemical
imaging.
MALDI and GALDI MS and MSn of total brain lipid extract at different sample
chamber pressures. Total brain lipid extract was studied under the various ionization modes
because the components and their structures have been relatively well identified or analyzed
by several mass spectral techniques. Mass spectral profiles of the brain lipid extract in
various ionization environments are shown in Figure 4. First, mass peaks from m/z 730 to m/z
810 are dominant in MALDI, whereas mass peaks from m/z 820 to m/z 880 are the major
peaks in GALDI. This tendency is similar to that in the PC/GlcCer lipid standard mixtures.
Second, for MALDI, no major difference between HV and IP was observed (Figure 4(a)
versus 4(b)). Third, for GALDI, the higher mass peak intensities in the region of m/z 820-880
and the lower mass peak intensities in the region of m/z 730-810 are more pronounced in HV
than in IP (Figure 4(c) versus 4(d)). In other words, lipid species in the region of m/z 820-
880 were detected more selectively in HV than in IP. This trend is also analogous to that in
the mass profiles of PC/GlcCer lipid standard mixtures.
To identify the lipid species detected, MS/MS and MS3 experiments for most of the
major peaks the mass spectra of brain lipid extract (Figure 4(b) and (d)) were performed.
Many structural studies of cerebrosides using tandem mass spectrometry have been reported
with ESI MS,3, 10, 77 but not many with LDI MS so far. For example, GALDI 1st generation
product ion spectrum of m/z 850.66, the most intense peak in Figure 4(d), is shown in Figure
5(a). Neutral loss of 162 Da (NL 162) and NL 180 correspond to the loss of C6H10O5 and the
loss of a galactose unit at the head group position of the sphingosine long chain. NL 18,
which corresponds to the loss of water, was also observed. The predominant mass peak at
m/z 484.50 corresponds to the sodiated C18 sphingosine long chain base and m/z 512 can be
assigned as the sodiated aldehyde form of the C18 sphingosine long chain base (see the
scheme on Figure 5(a)). The mass peak at m/z 850.66 can be assigned as [GalCer
25
d18:1/24:0h + Na]+. GalCer d18:1/24:0h has been known to be the major hydroxylated
cerebroside compound in brain.3, 78 Both NL 162 and NL 180 are highly characteristic
patterns in the first generation product ion spectra for precursor ions of cerebrosides. For all
mass peaks assigned as cerebrosides including both standards (data not shown) and the brain
lipid extract, these two NLs were always observed in their first generation product ion spectra.
In addition, the mass peaks at m/z 484 (C18 sphingosine long chain base) and at m/z 486 (C18
sphinganine long chain base) represent also significant mass fragment ions for cerebrosides
and one of both was observed in most cerebrosides (Table 3). To verify the structure of the
cerebroside further, MS3 experiments was also performed for the m/z 484.50 ion (Figure
5(b)). NL 162 and NL 180 were observed again in the second generation product ion mass
spectrum. In addition to these, the loss of NH3 (NL 17) was also observed. The peaks at m/z
203 and m/z 185 could be identified as the sodiated galactose ion and subsequent loss of
water from it.
Mass peak assignments for cerebrosides and PCs based on the major fragment ions
and NLs are listed in Table 3. Only the major ions which have been verified by first
generation product ion spectra are listed. Fragment ions and neutral loss patterns of
cerebroside species have already been explained with Figure 5. For PCs, the characteristic
mass fragment ion m/z 184 (phosphocholine head group), was observed for some PC
precursor ions. NL 18(-H2O), NL 59
(–N(CH3)3), NL 124(-ethyl phosphate), NL 183(-phosphocholine head group) were observed
for PCs. Similar patterns were observed in earlier studies75, 79-81 and the information was
used for identifying PCs. NL 146, NL 162, NL 205 were also observed for some PC
precursor ions. NL 146 could correspond to NL of ethyl phosphate with sodium (ethyl
phosphate + Na – H) and NL 205 could correspond to NL of phosphocholine head group
with sodium (phosphocholine head group + Na – H). NL 162 from PC precursor ions could
correspond to NL of ethyl phosphate with potassium (ethyl phosphate + K – H). Note that
26
there are two neutral losses which have the same decrease of 162 Da but originated from
different species; NL 162 from GalCer precursor ions and NL 162 from adduct ions
consisting of PC and potassium. The difference between these two neutral losses is that NL
162 from GalCers precursor ions was always observed with NL 180 (-galactose) but NL 162
from PC precursor ions was not. In Table 3, NL 162 from PC precursor ions was annotated
as “NL 162k” to distinguish from NL 162 from GalCer precursor ions. As expected from the
analysis of the standard lipid mixture, GALDI gives more informative mass profiles for
cerebroside species whereas MALDI gives more informative mass profiles for PCs.
In Table 3, note that more than one lipid species can be associated with one m/z.
There are two possible causes for these isobaric ions in mass spectrum. First, several types of
adduct ions can be formed, even if two compounds have quite different monoisotopic masses.
In the case of PC 38:4 and PC 36:1, which are present in the brain, their difference in
calculated exact mass is 21.98 Da. This is the same as the calculated exact mass difference
between H and Na. Therefore, [PC 38:4 + H]+ and [PC 36:1 + Na]+ can be detected
simultaneously at m/z 810. Second, if the monoisotopic masses of two or more compounds
are too close to each other (less than 0.1 Da difference), they overlap and cannot be selected
separately for MS/MS experiments. For example, the monoisotopic masses for GalCer
d18:1/24:1 and PC 38:4 are 809.59 Da and 809.67 Da respectively and both compounds are
known to be present in the brain. They were observed together in both MALDI and GALDI
as listed in Table 3.
However, the ratio of [GalCer d18:1/24:1 + Na]+ and [PC 38:4 + Na]+ in MALDI MS
was different from that in GALDI MS. Figure 6 shows the two first generation product ion
spectra of m/z 832. Similar fragment patterns were observed but the relative abundances are
different in the two modes. The mass peak at m/z 773.25 corresponding to NL 59 from PCs is
dominant in IP-MALDI product ion spectrum (Figure 6(a)) while the mass peak at m/z
670.58 corresponding to NL 162 from cerebrosides is the most intense peak in IP-GALDI
27
product ion spectrum (Figure 6(b)). Comparison between the two mass spectra confirms that
GALDI MS is advantageous when cerebrosides need to be analyzed without separation from
the lipid pools, such as crude lipid extracts or tissue samples. In both mass spectra, mass
peaks corresponding NL 43 and NL 87 were also observed. NL 87 could correspond to the
loss of the C3H5O2N from the phosphatidylserine head group82 and NL 43 could be the
neutral loss of aziridine from phosphatidylethanolamine.83 This means that many other lipids,
such as PE, PS, and PG, can also be incorporated into one m/z for the same reasons.
Detection of sulfatides in negative ion mode by GALDI MS. Sulfatides (STs), one
of the derivatives of GalCers, are 3-sulfate esters of GalCers. STs are enriched in brain
tissues and mediate several biological processes including cell growth, cell adhesion, and
morphogenesis.84 By using MALDI MS, STs were successfully detected from brain tissue,
serum and renal tubule cells in earlier studies.19, 85, 86 Because ST has a sulfate group (SO4–),
they are also easily detected in negative ion mode as shown in Figure 7. The hydroxylated
forms of STs were also observed. Since C24 fatty acids are dominant in the mammalian
nervous system,84 STs and hydroxylated STs which have 24:1 or 24:0 fatty acid composition
were observed as the major species here.
IMS and IMS/MS in rat brain tissues by using GALDI MS. To analyze
constituents directly from tissue samples, ionization assisting materials need to be coated on
top of the tissue. Because extract or other liquid samples were loaded on top of the graphite
layer, the thickness of the graphite layer was not critical in extract analysis. In tissue analysis,
however, too small an amount of colloidal graphite could give weak signals and too large an
amount could generate strong interference signals from the colloidal graphite. To spray more
reproducibly, an airbrush was used with the four-times diluted colloidal graphite solution
instead of an aerosol spray can. We note that too dilute (over ten times) a solution could not
give a firmly-coated surface.
28
A lipid mass profile in the positive ion mode was generated from a rat brain tissue as
shown in Figure 8(b). The lipid profile in negative-ion mode is shown in Supporting
Information Figure S-5. To identify lipid species, MS/MS and MS3 experiments were
performed for major peaks in the lipid profile mass spectrum. For example, the mass peak at
m/z 844.50 in Figure 8(b) could be identified as [PC 38:6 +K]+ and [PE 38:4 + 2K – H]+
based on previous studies72, 83 and first generation product ion mass spectrum (See
Supporting Information Figure S-6). Chemically-selective images for the identified ions are
shown in Figure 9. Note that not only the ions listed in Figure 9 but also other isobaric ions
could contribute to these chemically selective images as discussed previously. From the
images in Figure 9, cerebroside-rich areas and cerebroside-deficient areas can be recognized
clearly and the extracted mass spectra from these two regions are shown in Figure 8(c) and
8(d). The mass region of m/z 800-880, where most of the cerebrosides were found, is
obviously different in Figures 8(c) and 8(d) in terms of mass profiles and relative intensities.
To separate the spatial information of isobaric ions from each other, IMS/MS was
performed. The mass peak at m/z 832.58 was chosen here because both PC and cerebroside
species show this common mass peak from the lipid extract analysis. First product ion
spectra of ions at m/z 832.58 (data are not shown) from the rat brain tissue have the same
fragmentation patterns as those from lipid extracts (Figure 6). Chemically-selective images of
the precursor ion at m/z 832, the product ion at m/z 773, and the product ion at m/z 670 are
shown in Figure 10. The spatial distribution of product ions at m/z 670, which corresponds to
NL 162 from [GalCer d18:1/24:1 + Na]+, is similar to those of other GalCer species shown in
Figure 9. Figure 10 shows that the spatial distribution of the two isobaric ions can be clearly
separated by performing IMS/MS.
29
Conclusions
In this study, a simple and effective method of LDI based on colloidal graphite was
developed and was applied to lipid analyses. Comparison between GALDI and MALDI of
standard lipid mixtures clearly showed the increase in signal from cerebrosides for the former,
making the former more suitable for detection of cerebroside species from lipid pools
without separation. With total brain lipid extract, twenty-two cerebroside species were
detected as major compounds by IP-GALDI MS and confirmed by MSn experiments while
only eight cerebroside species were verified by IP-MALDI MS and MSn experiments. From
Table 3 and Figure 4(c), we can conclude that the major mass peaks in HV-GALDI mass
spectrum correspond to cerebroside species. The peak enhancement for cerebrosides and the
peak suppression for PCs were more significant in HV than in IP. Therefore, HV could be
more suitable for selective detection of cerebrosides and IP could be more appropriate for
profiling various kinds of lipid species in complex lipid mixtures. Negative ion mode of
GALDI also showed comparable performance as MALDI for detecting sulfatides in brain
lipid extracts. For rat brain tissue, mass profiles and chemically-selective images for lipids
were successfully obtained with the modified colloidal graphite spraying technique.
Cerebrosides and sulfatides were well detected in tissue analysis as in extract analysis, and
the increase in intensities of cerebrosides facilitated the generation of images for cerebrosides.
Automated MS/MS experiments over the tissue surface and image processing from these
experiments showed the inherent problem of full mass scan data and the need for MS/MS
imaging. However, performing MS/MS for even one precursor ion over the tissue surface is
already time consuming. Therefore, faster protocols for IMS and further development of
IMS/MS are needed. In this work, the issue about sensitivity increase for fragile
biomolecules in IP LDI MS due to stabilization of labile bonds by collisional cooling was not
discussed. Future studies on the analyses of labile molecules such as gangliosides in HV and
IP by using GALDI MS and MSn would be valuable.
30
Acknowledgements
E.S.Y. thanks the Robert Allen Wright Endowment for Excellence for support. The
Ames Laboratory is operated for the US Department of Energy by Iowa State University
under Contract No. W-7405-Eng-82. This work was supported by the Director of Science,
Office of Basic Energy Sciences, Divisions of Chemical Sciences and Biosciences.
Appendixes
Supplemental GALDI mass spectra for lipid mixture standards with various ratios and
rat brain tissues, and product ion mass spectrum for the m/z 844 are listed in Appendix 1. In
Appendix 2, a complete scientific paper which describes GALDI MS studies of fruit samples
is presented.
Table 1. Mass peak assignments with relative intensities for lipid species in lipid standard mixturesa under different sample chamber pressures
Identification
High-Vacuum Intermediate Pressure
Theoretical m/z (Da)b
Observed m/z (Relative
Intensityc) with DHB
Observed m/z (Relative Intensity)
with Graphite
Observed m/z (Relative Intensity) with DHB
Observed m/z (Relative Intensity)
with Graphite
[GlcCer 16:0+Na]+ d 722.55 722.59 (29.4) 722.50 (78.2) 722.66 (12.3) 722.58 (54.8) [PC 32:0+H]+ e 734.57 734.56 (100.0) — 734.50 (100.0) 734.50 (17.6) [GlcCer 18:0+Na]+ 750.58 750.52 (4.4) 750.53 (13.0) 750.66 (2.6) 750.66 (12.8) [PC 32:0+Na]+ 756.55 756.52 (39.7) 756.57 (6.8) 756.50 (81.2) 756.50 (81.2) [PC 32:0+K]+ 772.53 — — — 772.50 (7.8) [GlcCer 20:0+Na]+ 778.62 778.56 (10.1) 778.59 (30.5) 778.66 (5.0) 778.66 (27.3) [GlcCer 21:0+Na]+ 792.63 792.62 (11.0) 792.60 (38.6) 792.66 (5.6) 792.66 (33.7) [GlcCer 22:0+Na]+ 806.65 806.66 (23.6) 806.63 (100.0) 806.66 (16.5) 806.66 (100.0) [GlcCer 23:0+Na]+ 820.66 820.60 (19.6) 820.65 (67.2) 820.66 (13.2) 820.66 (85.1) [GlcCer 24:0+Na]+ 834.68 834.61 (13.1) 834.66 (41.4) 834.75 (8.2) 834.66 (53.8)
aStandard lipid mixture contains 0.5mg/ml of PC 32:0 and 0.5mg/ml of total GlcCers. bMasses are monoisotopic. cRelative intensity = [(Intensity of relevant mass peak)/(Intensity of the most intense mass peak in the mass spectrum)] × 100. dAll glucosylceramide standards have C18 sphingosine (d18:1) base and the numbers m:n correspond to the number of the total carbons (m):number of double bonds (n) in the amide-linked fatty acid. eNumber x:y corresponds to the number of the total carbons of fatty acids at sn-1 and sn-2 position (x): number of double bonds (y) of PC.
31
Table 2. Relative peak intensities of major peaks in lipid standard mixtures with various PC/GlcCers ratiosa Concentration (mg/ml) Relative Intensity with DHB (%) Relative Intensity with Graphite (%)
PC 32:0 GlcCers [PC32:0+H]+ [GlcCer22:0+Na]+ [PC32:0+Na]+ [GlcCer22:0+Na]+
Pressure HV IP HV IP HV IP HV IP
Batch 1
0.25
0.5
346.6 559.3
100
12.4 57.1
100 0.125 186.8 311.6 10.1 33.2 0.025 115.5 134.1 n.d. 7.6 0.005 83.9 112.8 n.d. 5.1
Batch 2 0.5
0.25
100
9.2 8.3
100
634.5 95.0
0.125 8.4 6.0 410.3 33.0
0.025 2.2 4.5 30.5 15.7
0.005 n.d. 3.4 n.d. 8.3
aMass spectra used for calculation are listed in Supporting Information Figure S-1, S-2, S-3, and S-4. bn.d. represents “not detected”.
32
33
Table 3. Mass peak assignments for cerebrosides and phosphatidylcholines in brain lipid extract based on MS/MS experiments
IP-GALDI IP-MALDI Observed Product Ionsa
and Neutral Losses (NLs)b Ions Assigned m/z Ions Assigned Observed Product Ions
and Neutral Losses (NLs) Cerebrosides Cerebrosides
NL 18, NL 162, NL 180 [GalCer d18:1/18:0c
+ Na]+
750
m/z 484, NL 18, NL 162, NL 180 [GalCer d18:1/18:0h
d+
Na]+
766
NL 18, NL 162, NL 180 [GalCer d18:1/22:1 + Na]+
804
NL 18, NL 162, NL 180 [GalCer d18:1/22:0 + Na]+
806 [GalCer d18:1/22:0 + Na]+
NL 18, NL 162, NL180
m/z 484, NL 18, NL 162, NL 180 [GalCer d18:0/22:0 + Na]+
808 [GalCer d18:0/22:0 + Na]+
m/z 484, NL 18, NL 162, NL180
m/z 484, m/z 512, NL 18, NL 162, NL 180 [GalCer d18:1/23:0 + Na]+
820
m/z 484, m/z 512, NL 18, NL 162, NL 180 [GalCer d18:1/22:0h + Na]+
822 [GalCer d18:1/22:0h + Na]+
m/z 484, m/z 512, NL 18, NL 162, NL180
m/z 486, m/z 514, NL 18, NL 162, NL 180 [GalCer d18:0/22:0h + Na]+
824
NL 18, NL 162, NL 180 [GalCer d18:1/24:1 + Na]+
832 [GalCer d18:1/24:1 + Na]+
NL 18, NL 162, NL180
m/z 484, m/z 512, NL 18, NL 162, NL 180 [GalCer d18:1/24:0 + Na]+
834
m/z 484, m/z 512, NL 18, NL 162, NL 180 [GalCer d18:1/23:0h + Na]+
836 [GalCer d18:1/23:0h + Na]+
m/z 484, NL 18, NL 162, NL180
m/z 486, NL 18, NL 162, NL 180 [GalCer d18:0/23:0h + Na]+
838
m/z 484, m/z 512, NL 18, NL 162, NL 180 [GalCer d18:1/24:1h + Na]
+
[GalCer d18:1/25:0 + Na]+
848
[GalCer d18:1/24:1h + Na]+
[GalCer d18:1/25:0 + Na]+
m/z 484, m/z 512, NL 18, NL 162, NL180
m/z 484, m/z 512, NL 18, NL 162, NL 180 [GalCer d18:1/24:0h + Na]+
850 [GalCer d18:1/24:0h+Na]+
m/z 484, m/z 512, NL 18, NL 162, NL180
m/z 486, m/z 514, NL 18, NL 162, NL 180 [GalCer d18:0/24:0h + Na]+
852 [GalCer d18:0/24:0h+Na]+
m/z 486, NL 18, NL 162, NL180
m/z 484, NL 18, NL 162, NL 180 [GalCer d18:1/26:1 + Na]+
860
m/z 484, m/z 512, NL 18, NL 162, NL 180 [GalCer d18:1/25:1h + Na]+
862
m/z 484, m/z 512, NL 18, NL 162, NL 180 [GalCer d18:1/25:0h + Na]+
864
m/z 486, NL 18, NL 162, NL 180 [GalCer d18:0/25:0h + Na]+
866
m/z 484, m/z 512, NL 18, NL 162, NL 180 [GalCer d18:1/26:1h + Na]+
876
m/z 484, m/z 512, NL 18, NL 162, NL 180 [GalCer d18:1/26:0h + Na]+
878
m/z 486, m/z 514, NL 18, NL 162, NL 180 [GalCer d18:0/26:0h + Na]+
880 Phosphatidylcholines Phosphatidylcholines
NL 124, NL 146 [PC 34:1 - N(CH3)3 + Na]+
723
734 [PC 32:0 + H]+
m/z 184, NL 59, NL124
NL 124, NL 162k [PC 34:1 - N(CH3)3 + K]+
739
NL 124, NL 146 [PC 36:1 - N(CH3)3 + Na]+
751
NL 59, NL 124, NL146 [PC 32:0 + Na]+
756 [PC 32:0 + Na]+
NL 59, NL 183, NL 205
760 [PC 34:1 + H]+
m/z 184, NL 59, NL 124, NL183
762 [PC 34:0 + H]+
m/z 184, NL 59
NL 59 [PC 32:0 + K]+
772 [PC 32:0 + K]+
NL 59, NL183
NL 59 [PC 34:1 + Na]+
782 [PC 34:1 + Na]+
NL 59, NL 124, NL 183, NL 205
784 [PC 34:0 + Na]+
NL 59, NL 183, NL 205
786 [PC 36:2 + H]+
NL 59, NL 183
788 [PC 36:1 + H]+
NL 59, NL183
NL 59, NL162k [PC 34:1 + K]+
798 [PC34:1 + K]+
NL 59, NL 183
806 [PC 38:6 + H]+
NL 59, NL 124, NL 183
NL 59 [PC 36:2 + Na]+
808 [PC 36:2 + Na]+
NL 59, NL 124, NL 183
NL 59 [PC 36:1 + Na]+
810 [PC 36:1 + Na]
+
a/o [PC 38:4 + H]+
NL 59, NL 124
NL 59 [PC 38:4 + Na]+
832 [PC 38:4 + Na]+
NL 59, NL 124, NL 183
836 [PC 38:2 + Na]+
NL 59, NL 183 a,bSee text for assignments of product ions and neutral losses cGalCer A/B corresponds to galactosylceramide sphingoid long chain base (A)/amide-linked fatty acid (B). dh corresponds to the hydroxyl group at the C2 position of the amide-linked fatty acid.
34
References
(1) Bektas, M.; Spiegel, S. Glycoconjugate Journal 2003, 20, 39-47.
(2) Tan, R. X.; Chen, J. H. Natural Product Reports 2003, 20, 509-534.
(3) Han, X. L.; Cheng, H. Journal of Lipid Research 2005, 46, 163-175.
(4) Lingwood, C. A. Current Opinion in Chemical Biology 1998, 2, 695-700.
(5) Coetzee, T.; Fujita, N.; Dupree, J.; Shi, R.; Blight, A.; Suzuki, K.; Suzuki, K.; Popko,
B. Cell 1996, 86, 209-219.
(6) Scriver, C. R., 8th ed.; McGraw-Hill: New York, 2001, pp 2641-2692.
(7) Han, X. L.; Holtzman, D. M.; McKeel, D. W.; Kelley, J.; Morris, J. C. Journal of
Neurochemistry 2002, 82, 809-818.
(8) Han, X. L.; Cheng, H.; Fryer, J. D.; Fagan, A. M.; Holtzman, D. M. Journal of
Biological Chemistry 2003, 278, 8043-8051.
(9) Hauser, G. Journal of Neurochemistry 1968, 15, 1237-1238.
(10) Han, X. L.; Gross, R. W. Mass Spectrometry Reviews 2005, 24, 367-412.
(11) Whitehouse, C. M.; Dreyer, R. N.; Yamashita, M.; Fenn, J. B. Analytical Chemistry
1985, 57, 675-679.
(12) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T.; Matsuo, T. Rapid
Communications in Mass Spectrometry 1988, 2, 151-153.
(13) Karas, M.; Hillenkamp, F. Analytical Chemistry 1988, 60, 2299-2301.
(14) Schiller, J.; Suss, R.; Arnhold, J.; Fuchs, B.; Lessig, J.; Muller, M.; Petkovic, M.;
Spalteholz, H.; Zschornig, O.; Arnold, K. Progress in Lipid Research 2004, 43, 449-
488.
(15) Welti, R.; Wang, X. M. Current Opinion in Plant Biology 2004, 7, 337-344.
(16) Petkovic, M.; Schiller, J.; Muller, M.; Benard, S.; Reichl, S.; Arnold, K.; Arnhold, J.
Analytical Biochemistry 2001, 289, 202-216.
(17) Harvey, D. J. Journal of Mass Spectrometry 1995, 30, 1333-1346.
35
(18) Harvey, D. J. Journal of Mass Spectrometry 1995, 30, 1311-1324.
(19) Jackson, S. N.; Wang, H. Y. J.; Woods, A. S. Analytical Chemistry 2005, 77, 4523-
4527.
(20) Estrada, R.; Yappert, M. C. Journal of Mass Spectrometry 2004, 39, 412-422.
(21) Schiller, J.; Subb, R.; Petkovic, M.; Hilbert, N.; Muller, M.; Szchornig, O.; Arnhold,
J.; Arnold, K. Chemistry and Physics of Lipids 2001, 113, 123-131.
(22) Juhasz, P.; Costello, C. E. Journal of the American Society for Mass Spectrometry
1992, 3, 785-796.
(23) Fujiwaki, T.; Yamaguchi, S.; Sukegawa, K.; Taketomi, T. Journal of
Chromatography B 1999, 731, 45-52.
(24) O'Connor, P. B.; Costello, C. E. Rapid Communications in Mass Spectrometry 2001,
15, 1862-1868.
(25) O'Connor, P. B.; Mirgorodskaya, E.; Costello, C. E. Journal of the American Society
for Mass Spectrometry 2002, 13, 402-407.
(26) Ivleva, V. B.; Elkin, Y. N.; Budnik, B. A.; Moyer, S. C.; O'Connor, P. B.; Costello, C.
E. Analytical Chemistry 2004, 76, 6484-6491.
(27) Dreisewerd, K.; Muthing, J.; Rohlfing, A.; Meisen, I.; Vukelic, Z.; Peter-Katalinic, J.;
Hillenkamp, F.; Berkenkamp, S. Analytical Chemistry 2005, 77, 4098-4107.
(28) Ivleva, V. B.; Sapp, L. M.; O'Connor, P. B.; Costello, C. E. Journal of the American
Society for Mass Spectrometry 2005, 16, 1552-1560.
(29) Taketomi, T.; Sugiyama, E.; Uemura, K.; Hara, A.; Hidaka, H.; Tozuka, M.;
Nakabayashi, T.; Katsuyama, T. Journal of Lipid Research 2001, 42, 873-885.
(30) Yurkova, I.; Kisel, M.; Arnhold, J.; Shadyro, O. Chemistry and Physics of Lipids
2005, 134, 41-49.
(31) Fujiwaki, T.; Yamaguchi, S.; Tasaka, M.; Sakura, N.; Taketomi, T. Journal of
Chromatography B-Analytical Technologies in the Biomedical and Life Sciences
36
2002, 776, 115-123.
(32) Sunner, J.; Dratz, E.; Chen, Y. C. Analytical Chemistry 1995, 67, 4335-4342.
(33) Dale, M. J.; Knochenmuss, R.; Zenobi, R. Analytical Chemistry 1996, 68, 3321-3329.
(34) Peng, S.; Edler, M.; Ahlmann, N.; Hoffmann, T.; Franzke, J. Rapid Communications
in Mass Spectrometry 2005, 19, 2789-2793.
(35) Zumbuhl, S.; Knochenmuss, R.; Wulfert, S.; Dubois, F.; Dale, M. J.; Zenobi, R.
Analytical Chemistry 1998, 70, 707-715.
(36) Kim, J.; Kang, W. Bulletin of the Korean Chemical Society 2000, 21, 401-404.
(37) Kim, J.; Paek, K.; Kang, W. Bulletin of the Korean Chemical Society 2002, 23, 315-
319.
(38) Kim, H. J.; Lee, J. K.; Park, S. J.; Ro, H. W.; Yoo, D. Y.; Yoon, D. Y. Analytical
Chemistry 2000, 72, 5673-5678.
(39) Park, K. H.; Kim, H. J. Rapid Communications in Mass Spectrometry 2001, 15, 1494-
1499.
(40) Li, X. P.; Wilm, M.; Franz, T. Proteomics 2005, 5, 1460-1471.
(41) Black, C.; Poile, C.; Langley, J.; Herniman, J. Rapid Communications in Mass
Spectrometry 2006, 20, 1053-1060.
(42) Chen, Y. C.; Shiea, J.; Sunner, J. Journal of Chromatography A 1998, 826, 77-86.
(43) Ugarov, M. V.; Egan, T.; Khabashesku, D. V.; Schultz, J. A.; Peng, H. Q.;
Khabashesku, V. N.; Furutani, H.; Prather, K. S.; Wang, H. W. J.; Jackson, S. N.;
Woods, A. S. Analytical Chemistry 2004, 76, 6734-6742.
(44) Ren, S. F.; Zhang, L.; Cheng, Z. H.; Guo, Y. L. Journal of the American Society for
Mass Spectrometry 2005, 16, 333-339.
(45) Xu, S. Y.; Li, Y. F.; Zou, H. F.; Qiu, J. S.; Guo, Z.; Guo, B. C. Analytical Chemistry
2003, 75, 6191-6195.
(46) Ren, S. F.; Guo, Y. L. Rapid Communications in Mass Spectrometry 2005, 19, 255-
37
260.
(47) Pan, C. S.; Xu, S. Y.; Hu, L. G.; Su, X. Y.; Ou, J. J.; Zou, H. F.; Guo, Z.; Zhang, Y.;
Guo, B. C. Journal of the American Society for Mass Spectrometry 2005, 16, 883-892.
(48) Wu, K. J.; Odom, R. W. Analytical Chemistry 1996, 68, 873-882.
(49) Delcorte, A.; Bour, J.; Aubriet, F.; Muller, J. F.; Bertrand, P. Analytical Chemistry
2003, 75, 6875-6885.
(50) Delcorte, A.; Medard, N.; Bertrand, P. Analytical Chemistry 2002, 74, 4955-4968.
(51) Sjovall, P.; Lausmaa, J.; Johansson, B. Analytical Chemistry 2004, 76, 4271-4278.
(52) Heeren, R. M. A.; McDonnell, L. A.; Amstalden, E.; Luxembourg, S. L.; Altelaar, A.
F. M.; Piersma, S. R. Applied Surface Science 2006, 252, 6827-6835.
(53) Caprioli, R. M.; Farmer, T. B.; Gile, J. Analytical Chemistry 1997, 69, 4751-4760.
(54) Chaurand, P.; DaGue, B. B.; Pearsall, R. S.; Threadgill, D. W.; Caprioli, R. M.
Proteomics 2001, 1, 1320-1326.
(55) Masumori, N.; Thomas, T. Z.; Chaurand, P.; Case, T.; Paul, M.; Kasper, S.; Caprioli,
R. M.; Tsukamoto, T.; Shappell, S. B.; Matusik, R. J. Cancer Research 2001, 61,
2239-2249.
(56) Schwartz, S. A.; Weil, R. J.; Johnson, M. D.; Toms, S. A.; Caprioli, R. M. Clinical
Cancer Research 2004, 10, 981-987.
(57) Chaurand, P.; Cornett, D. S.; Caprioli, R. M. Current Opinion in Biotechnology 2006,
17, 431-436.
(58) Chaurand, P.; Caprioli, R. M. Electrophoresis 2002, 23, 3125-3135.
(59) Spengler, B.; Hubert, M. Journal of the American Society for Mass Spectrometry
2002, 13, 735-748.
(60) Luxembourg, S. L.; McDonnell, L. A.; Mize, T. H.; Heeren, R. M. A. Journal of
Proteome Research 2005, 4, 671-673.
(61) Luxembourg, S. L.; Mize, T. H.; McDonnell, L. A.; Heeren, R. M. A. Analytical
38
Chemistry 2004, 76, 5339-5344.
(62) Luxembourg, S. L.; McDonnell, L. A.; Duursma, M. C.; Guo, X. H.; Heeren, R. M. A.
Analytical Chemistry 2003, 75, 2333-2341.
(63) Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. Journal of Mass Spectrometry 2003,
38, 699-708.
(64) Stoeckli, M.; Staab, D.; Staufenbiel, M.; Wiederhold, K. H.; Signor, L. Analytical
Biochemistry 2002, 311, 33-39.
(65) Bogan, M. J.; Agnes, G. R. Rapid Communications in Mass Spectrometry 2004, 18,
2673-2681.
(66) Aerni, H. R.; Cornett, D. S.; Caprioli, R. M. Analytical Chemistry 2006, 78, 827-834.
(67) Armstrong, D. W.; Zhang, L. K.; He, L. F.; Gross, M. L. Analytical Chemistry 2001,
73, 3679-3686.
(68) Tholey, A.; Heinzle, E. Analytical and Bioanalytical Chemistry 2006, 386, 24-37.
(69) Tholey, A. Rapid Communications in Mass Spectrometry 2006, 20, 1761-1768.
(70) Mank, M.; Stahl, B.; Boehm, G. Analytical Chemistry 2004, 76, 2938-2950.
(71) Lemaire, R.; Tabet, J. C.; Ducoroy, P.; Hendra, J. B.; Salzet, M.; Fournier, I.
Analytical Chemistry 2006, 78, 809-819.
(72) Garrett, T. J.; Prieto-Conaway, M. C.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford,
G.; Yost, R. A. Int. J. Mass Spectrom. 2006, In Press,
doi:10.1016/j.ijms.2006.1009.1019.
(73) Garden, R. W.; Sweedler, J. V. Analytical Chemistry 2000, 72, 30-36.
(74) Hanton, S. D.; Clark, P. A. C.; Owens, K. G. Journal of the American Society for
Mass Spectrometry 1999, 10, 104-111.
(75) Garrett, T. J.; Yost, R. A. Analytical Chemistry 2006, 78, 2465-2469.
(76) Garrett, T. J.; Prieto-Conaway, M. C.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford,
G.; Yost, R. A. Int. J. Mass Spectrom. 2006, doi:10.1016/j.ijms.2006.09.019.
39
(77) Hsu, F. F.; Turk, J. Journal of the American Society for Mass Spectrometry 2001, 12,
61-79.
(78) Kaye, E. M.; Ullman, M. D. Analytical Biochemistry 1984, 138, 380-385.
(79) Ham, B. M.; Jacob, J. T.; Cole, R. B. Analytical Chemistry 2005, 77, 4439-4447.
(80) Hsu, F. F.; Turk, J. Journal of the American Society for Mass Spectrometry 2003, 14,
352-363.
(81) Al-Saad, K. A.; Zabrouskov, V.; Siems, W. F.; Knowles, N. R.; Hannan, R. M.; Hill,
H. H. Rapid Communications in Mass Spectrometry 2003, 17, 87-96.
(82) Hsu, F. F.; Turk, J. Journal of the American Society for Mass Spectrometry 2005, 16,
1510-1522.
(83) Hsu, F. F.; Turk, J. Journal of Mass Spectrometry 2000, 35, 596-606.
(84) Ishizuka, I. Progress in Lipid Research 1997, 36, 245-319.
(85) Kyogashima, M.; Tamiya-Koizumi, K.; Ehara, T.; Li, G.; Hu, R.; Hara, A.; Aoyama,
T.; Kannagi, R. Glycobiology 2006, 16, 719-728.
(86) Sugiyama, E.; Hara, A.; Uemura, K. Analytical Biochemistry 1999, 274, 90-97.
40
Figure Captions
Figure 1. Scanning electron microscopy image of the colloidal graphite sprayed surface.
Figure 2. LDI mass spectra of the standard lipid mixture under four different conditions,
(a) high-vacuum (HV)-MALDI, (b) intermediate pressure (IP)-MALDI, (c)
HV-GALDI, and (d) IP-GALDI. The standard lipid mixture contains
0.5mg/ml PC 32:0 and 0.5 mg/ml total GlcCers. Mass peak assignments and
relative abundances are listed in Table 1.
Figure 3. (a) On the left is the optical image of the MALDI sample spot consisting of 1
µl of the standard lipid mixture (0.25 mg/ml of PC 32:0 and 0.5 mg/ml of total
GlcCers) and 1 µl of 50 mg/ml of DHB matrix solution. The sample spot size
is about 2.7 mm by 2.4 mm. Following are four chemically-selective images
corresponding to the normalized intensity map (mass peak intensity divided
by total ion current of each point) of the major lipid species. (b) IP-MALDI
spectrum extracted from the region of “big, needle-like crystals” at the
peripheral of the sample spot. (c) IP-MALDI spectrum extracted from the
region of “small crystals” at the interior of the sample spot.
Figure 4. LDI mass spectra of total brain lipid extract under different conditions,
(a) HV-MALDI, (b) IP-MALDI, (c) HV-GALDI, and (d) IP-GALDI in
positive ion mode. Mass peak assignments and corresponding MS/MS data
are listed in Table 3.
41
Figure 5. (A) First generation product ion spectrum of ions detected at m/z 850.66 from
IP-GALDI mass spectrum of total brain lipid extract (Figure 4(d)). The neutral
loss of 18 Da (NL 18) corresponds to the loss of water. NL 162 corresponds to
the loss of C6H10O2 from the sphingosine long chain. NL 180 is the loss of a
galactose unit. Mass peaks at m/z 484 and at m/z 512 correspond to the
sodiated C18 sphingosine long chain and its aldehyde form as seen in the
scheme. (b) First generation product ion spectrum of ions at m/z 484.50 which
are the first generation product ions of m/z 850.66. NL 17 corresponds to the
loss of NH3. NL 162 and NL 180 are the same as those in Figure 5(a). The
precursor ion at m/z 850.66 can be assigned as [GalCer d18:1/24:0h+Na]+.
Figure 6. First generation product ion spectra of ions at m/z 832 from (a) IP-MALDI
(Figure 4(c)) and (b) IP-GALDI (Figure 4(d)) mass spectra of the total brain
lipid extract. NLs of 18, 162, and 180 are the same as those in Figure 5. NL 59,
NL 124, and NL183 correspond to losses of trimethylamine (N(CH3)3), ethyl
phosphate from PC, and phosphocholine head group, respectively. NL 87
could correspond to the loss of the C3H5O2N from the phosphatidylserine
group. Note that NL 59 from PC is dominant in the MALDI mass spectrum
whereas NL 162 and NL 180 from cerebrosides are dominant in the GALDI
mass spectrum.
Figure 7. IP-GALDI mass spectrum of total brain lipid extract in the negative ion mode.
Sulfatides (STs) were mainly detected. ST m:n(h) represents sulfatide with
number of the total carbons (m):number of double bonds (n) in the amide-
linked fatty acid (hydroxylated).
42
Figure 8. (a) Chemically-selective image of ions at m/z 810 with grayscale gradation-
coded intensities. (b) The mass spectrum is the average of mass spectra at all
rastering points on the tissue (10,732 points) in the positive-ion mode. Each
mass spectrum in (c) and (d) was extracted from the small specific area (3 × 3
rastering points) as indicated in (a).
Figure 9. Chemically-selective images in the positive-ion and negative-ion modes.
Major ionic species identified by MS/MS and MS3 experiment were listed
together with the intensity ranges that correspond to the scale bar at the top
center of the figure.
Figure 10. Chemically-selective images of rat brain tissue for (a) the precursor ion at m/z
832, (b) the product ion at m/z 773 which represents NL 59 from PCs, and (c)
the product ion at 670 which represents NL 162 from cerebrosides. Therefore,
the isobaric ions at m/z 832 corresponding to [GalCer d18:1/24:1 + Na]+ and
[PC 38:4 + Na]+ in (a) can be separated into the image for [PC 38:4 + Na]+ in
(b) and the image for [GalCer d18:1/24:1 + Na]+ in (c) by performing
IMS/MS.
43
Figure 1.
44
680 700 720 740 760 780 800 820 840 860 880 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
680 700 720 740 760 780 800 820 840 860 880 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
(a)
[Glc
Cer
16:
0 +
Na]
+
[PC 32:0 + H]+
[PC
32:
0 +
Na]
+
[Glc
Cer
20:
0 +
Na]
+
[Glc
Cer
21:
0 +
Na]
+
[Glc
Cer
22:
0 +
Na]
+
[Glc
Cer
23:
0 +
Na]
+
[Glc
Cer
24:
0 +
Na]
+
HV-MALDI(b)
[Glc
Cer
16:
0 +
Na]
+
[PC 32:0 + H]+
[PC 32:0 + Na]+
[Glc
Cer
20:
0 +
Na]
+
[Glc
Cer
21:
0 +
Na]
+
[Glc
Cer
22:
0 +
Na]
+
[Glc
Cer
23:
0 +
Na]
+
[Glc
Cer
24:
0 +
Na]
+
IP-MALDI
6800
10
20
30
40
50
60
70
80
90
100
700 720 740 760 780 800 820 840 860 880 900
Rel
ativ
e A
bund
ance
(%)
m/z
(c)
680 700 720 740 760 780 800 820 840 860 880 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
(d)
[Glc
Cer
16:
0 +
Na]
+
[PC
32:
0 +
H]+
[PC 32:0 + Na]+
[Glc
Cer
20:
0 +
Na]
+
[Glc
Cer
21:
0 +
Na]
+
[GlcCer 22:0 + Na]+
[GlcCer 23:0 + Na]+
[GlcCer 24:0 + Na]+
[GlcCer 16:0 + Na]+
[PC
32:
0 +
Na]
+
[Glc
Cer
20:
0 +
Na]
+
[Glc
Cer
21:
0 +
Na]
+
[GlcCer 22:0 + Na]+
[Glc
Cer
23:
0+N
a]+
[Glc
Cer
24:
0 +
Na]
+
[Glc
Cer
18:
0 +
Na]
+
HV-GALDI
[PC
32:
0 - N
(CH 3
) 3+
Na]
+
[PC
32:
0 - N
(CH 3
) 3+
K]+
[Glc
Cer
18:
0 +
Na]
+
[PC
32:
0 +
K]+
IP-GALDI
Figure 2.
45
680 700 720 740 760 780 800 820 840 860 880 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
(b)
680 700 720 740 760 780 800 820 840 860 880 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
(c)
0.00 0.05 0.10
(a)
[PC 32:0 + H]+
[PC 32:0 + Na]+
[PC 32:0 + H]+
[PC 32:0 + Na]+
[GlcCer 22:0 + Na]+
[PC 32:0 + H]+ [GlcCer 22:0 + Na]+[PC 32:0 + Na]+ [GlcCer 23:0 + Na]+
Figure 3.
46
720 740 760 780 800 820 840 860 880 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
731.
6273
2.60
734.57
760.57
788.
62
782.57
794.
6379
8.52 80
6.58
810.
59
820.
66 822.
66 834.
68
850.
72720 740 760 780 800 820 840 860 880 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
720 740 760 780 800 820 840 860 880 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
720 740 760 780 800 820 840 860 880 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
731.
50
734.50
750.58
756.
50
760.50
768.50
782.50
788.
50
798.
5080
6.50
810.50
832.
58
848.
6685
0.75
766.
5376
8.55
806.
48
822.65
832.
6583
6.66
838.63
850.65
723.
58
751.
4175
3.58 76
3.41
769.
50 772.
50
779.50
788.
6679
0.58
798.
5880
4.50 81
0.58
820.
66
822.58
834.
66
850.66
864.
5886
6.58
(a) (b)
(c) (d)
HV-MALDI IP-MALDI
HV-GALDI IP-GALDI
756.
56
824.
6883
2.73
848.
64753.
57
753.
50
808.
50
822.
66
835.
58
874.
83
804.
48
828.
47
739.
46
749.
36
778.
4478
2.55 79
0.51
794.
60
848.
6685
2.67
866.6586
4.65
868.
67 782.
50
826.
50
836.
6683
2.50
848.
6685
2.58
876.
75
Figure 4.
47
300 400 500 600 700 800 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e Ab
unda
nce
(%)
484.50
512.42 670.58
688.50
763.33
832.58
(a)
300 400 500 600 700 800 9000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e Ab
unda
nce
(%)
(b)
203.83
322.67
467.42
Amide-linked fatty acid
NL 162
NL 180
NL18
185.92 304.75
484.50NL180
NL162
NL17
OOH
HOOH
O
OH
(CH2)12CH3NH
O(CH2)21CH3
OH
OH
Na+
m/z 484NL 162NL 180
Na+
m/z 512
OOH
HOOH
O
OH
(CH2)12CH3
NH2
OHNL 162NL 180
NL 17
Figure 5.
48
450 500 550 600 650 700 750 800 8500
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
652.
42 (N
L180
)
670.
67 (N
L 16
2)
773.25 (NL 59)
814.
50 (N
L 18
)83
2.58
649.
33 (N
L183
)
708.
42 (N
L 12
4)
745.
33 (N
L 87
)
788.
92 (N
L 43
)
(a)
IP-MALDI
450 500 550 600 650 700 750 800 8500
10
20
30
40
50
60
70
80
90
100
m/z
652.
50 (N
L 18
0)
670.58 (NL 162)
814.
50 (N
L 18
)
Rel
ativ
e A
bund
ance
(%)
832.
50
789.
17 (N
L 43
)77
3.25
(NL
59)
(b)IP-GALDI
Figure 6.
49
780 800 820 840 860 880 900 920 9400
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
[ST
18:0
- H
]- (80
6.50
)
[ST
18:0
h - H
]- (82
2.41
)
[ST
24:1
- H
]- (88
8.58
)
[ST
23:0
- H
]- (87
8.58
)
[ST
22:0
h - H
]- (86
2.58
)
[ST
20:0
h - H
]- (85
0.50
)
[ST
20:0
- H
]- (83
4.50
)
[ST
26:0
- H
]- (91
8.58
)[S
T 26
:1 -
H]- (
916.
58)
[ST
24:0
h - H
]- (90
6.50
)[S
T 24
:1h
- H]- (
904.
58)
[ST
24:0
- H
]-
(8
90.5
8)
[ST
260h
- H
]- (93
4.58
)[S
T 26
:1h
- H]- (
932.
50)
Figure 7.
50
720
740
760
780
800
820
840
860
880
900
0102030405060708090100
m/z
Relative Abundance (%)
751.
50
756.42
772.42782.
50
798.
50
822.58826.50
850.58
848.
66
868.
41
720
740
760
780
800
820
840
860
880
900
0102030405060708090100
m/z
Relative Abundance (%)
756.
58
772.
50782.
50
798.
49
810.50
826.50
844.50 848.50852.50
868.50
(c)
(d)
(a)
720
740
760
780
800
820
840
860
880
900
0102030405060708090100
m/z
Relative Abundance (%)
723.42
756.
50
772.
50782.
58
798.
50
806.50
828.50
844.50
868.
50
852.50848.58
810.10
(b)
810.50
832.58
Figure 8.
51
m/z
888
[ST
24:1
- H
]-
(0-1
2000
0)
m/z
890
[ST
24:0
- H
]-
(0-7
0000
)
m/z
885
[PI 3
8:4
- H]-
(0-1
7500
)
m/z
782
[PC
34:
1 +
Na]
+
(0-2
0000
0)
m/z
810
[PC
36:
1 +
Na]
+
(0-1
3000
0)
m/z
756
[PC
32:
0 +
Na]
+
(0-1
4000
0)
m/z
822
[Gal
Cer
d18
:1/2
2:0h
+ N
a]+
(0-1
3000
0)
m/z
850
[Gal
Cer
d18
:1/2
4:0h
+ N
a]+
(0-1
4000
0)
Opt
ical
Imag
e
Neg
ativ
e Io
n M
ode
Pos
itive
Ion
Mod
e
0%10
0%
Figure 9.
52
Product ion at m/z 773 (NL 59)
Product ion at m/z 670 (NL 162)
0% 100%
m/z 832
Figure 10.
53
CHAPTER 3. DIRECT PROFILING AND IMAGING OF PLANT
METABOLITES IN INTACT TISSUES BY USING COLLOIDAL
GRAPHITE-ASSISTED LASER DESORPTION IONIZATION MASS
SPECTROMETRY
A paper accepted for publication in The Plant Journal*
Sangwon Cha, Hui Zhang, Hilal I. Ilarslan, Eve Syrkin Wurtele, Libuse Brachova,
Basil J. Nikolau and Edward S. Yeung
Abstract
Laser desorption/ionization (LDI)-based imaging mass spectrometry (MS) has been
applied to several biological systems to obtain information about both the identities of the
major chemical species and their localization. Colloidal graphite-assisted LDI (GALDI) MS
imaging was introduced for imaging of small molecules such as phospholipids, cerebrosides,
oligosaccharides, flavonoids, and other secondary metabolites with high spatial homogeneity
due to the finely dispersed particles. Mass profiles and images of Arabidopsis thaliana were
recorded directly from various plant surfaces and cross-sections. The main targeted
metabolites were flavonoids and cuticular waxes, both of which are important in many
aspects of functional genomics, proteomics, and metabolomics. Mass spectral profiles
revealed tissue-specific accumulation of flavonoids in flowers and petals. In addition, many
other location-specific ions were observed. The location and the degree of light-induced
flavonoid accumulation in stem sections were successfully probed by GALDI MS.
___________________________________________________________________________
* Reprint with permission from The Plant Journal 2008,
doi: 10.1111/j.1365-313X.2008.03507.x, Copyright © 2008 Blackwell Publishing Ltd.
54
Introduction
Because low molecular weight metabolites exhibit a variety of different chemical
properties, many analytical methods have been used to assay their structures, abundances,
and locations 1, 2. Among these methods are nuclear magnetic resonance (NMR) spectroscopy 3, 4 for structural analysis of the metabolites, vibrational spectroscopy for metabolic
fingerprinting, and mass spectrometry (MS) for metabolite profiling 5. Gas chromatography-
MS (GC-MS) has been extensively used over the last 20 years 1, 6, 7 because it is a relatively
well-constructed methodology that has good reproducibility and wide applicability to various
classes of metabolites. Moreover, GC-MS has the ability to identify and quantify metabolites.
Since the development of electrospray ionization (ESI) techniques 8 and advances in tandem
MS technologies, liquid chromatography-MS (LC-MS ) is also becoming a workhorse with
high throughput and high sensitivity 9.
It is known that metabolic processes in plants are spatially non-uniform. This is
readily apparent via microscopic examination of plant organs that reveals considerable
heterogeneity in cell morphology and the distribution of pigments (e.g., chlorophylls,
anthocyanins) among these cells. At the biochemical level a very good example of this cell-
level asymmetry is that of C-4 photosynthesis, which requires a distinct distribution of
metabolic processes among mesophyll and the bundle-sheath cells, respectively, to
concentrate carbon dioxide at the active site of RuBISCO (ribulose–1,5-biphosphate
carboxylase/oxygenase). This discovery required the physical separation of the two cell types
before their proteome and/or metabolome could be interrogated. Thus, the development of
analytical techniques that can interrogate metabolites in intact tissue and with high-spatial
resolution would offer the ability to decipher metabolic processes at the cellular level.
Laser desorption/ionization (LDI) MS and matrix-assisted laser ionization/desorption
(MALDI) MS have not been widely used in plant metabolomics mainly because of
difficulties in generating ions from the relatively hydrophobic species in plant tissue.
55
However, the applicability of LDI MS to imaging plant metabolites, including direct
profiling of cuticular wax compounds 10, analysis of phosphatidylcholine 11, condensed
tannins in bark 12, and profiling of plant carotenoids 13 has been successfully demonstrated.
The advantage of LDI MS, over hyphenated techniques (GC-, LC-, and CE- MS) is the
ability to sample biochemicals directly without any extraction protocol. Therefore, the spatial
distribution of the biochemicals is retained.
In laser desorption, the sampling resolution is governed primarily by the laser spot
size. Consequently, the spatial resolution of imaging MS by LDI can be much smaller than
that achievable by traditional methods such as dissecting plant materials followed by
chemical extraction prior to MS -The effort to achieve smaller sampling size by laser
microdissection with traditional GC-MS methods has been reported recently 14. Location-
specific sampling has made it possible for LDI to be employed as a microscope, that is, for
“imaging MS” 15. Imaging MS collects mass spectra consecutively over the sample surface
and reconstructs mass spectral data as an image for each specific compound. In addition to
LDI, ion beam-induced ionization methods such as secondary ionization (SIMS) 16-21 and
beam-induced desorption electrospray ionization (DESI) 22, 23 have also been successfully
applied to imaging MS.
In addition to laser spot size, factors that limit the spatial resolution of LDI imaging
MS include sensitivity of the mass spectrometer, abundance of a metabolite, and
homogeneity of the material added to facilitate ionization. Matrixes in MALDI MS, which
require co-crystallization with sample components to be effective, can decrease the spatial
resolution because of their irregular crystallization patterns. To overcome this, researchers
have introduced alternative matrixes that can be homogeneously distributed on the sample
surfaces. Recently, we used colloidal graphite , a sub-micron flake-like material, as an
assisting material (GALDI MS) for imaging lipids from rat brain slices and also for imaging
secondary metabolites such as flavonoids and organic acids from fruits 24, 25. The other
56
advantages of GALDI MS are that graphite is more hydrophobic, making it compatible with
plant materials, and is a near-blackbody, making it suitable for irradiation with any
wavelength of light. Hence, GALDI MS makes possible to interrogate some classes of
compounds that are not detected by conventional MALDI. Recently, Siuzdak’s group
introduced nanostructure-initiator mass spectrometry (NIMS) where ‘initiator’ molecules
trapped in nanostructures help to ionize intact molecules on their surfaces 26. NIMS can
achieve high spatial resolution because of its structure and it can accept both photon and ion
irradiation.
SIMS has been used for imaging studies of metal ions on plant surfaces 27-32. Recently,
localization of phenolic compounds from wood stems has been examined by time-of-flight
(TOF) SIMS 33, 34. However, most studies by LDI MS imaging have been done on animal
tissue sections such as rat brain slices. Recently, MS imaging of amino acids, sugars, and
phosphorylated metabolites on wheat seeds by using conventional MALDI matrix was
reported 35. IR laser, instead of conventional UV laser, has been used to image several
abundant metabolites from fruits 36. There, the ultimate spatial resolution that can be
achieved is compromised because of the wavelength of light.
Elongation of aliphatic fatty acids to form very long chain fatty acids (VLCFAs) is
the first step in the biosynthesis of components of the cuticular wax layer. This layer makes
plants resistant to external stresses such as insects, pathogens, and water 37-41. Flavonoids
serve as epidermal shields to protect against UV light. In the case of Arabidopsis thaliana,
the major building blocks of flavonoid are flavonols and flavonol glycosides as shown in
Figure 1 42-44. Another major class of flavonoids is anthocyanins and proanthocyanidins, end
product of flavonoid’s biosynthetic pathways 45-47.
Because Arabidopsis is an important model system for plants, and as such the most
highly understood, it is important to be able to analyze its metabolome in situ. The
Arabidopsis plant is small and fragile; therefore its analysis is challenging for sample
57
handling and MS imaging. For the present imaging MS study, we focused on very long chain
fatty acids (VLCFAs) and the flavonoids because we have already demonstrated the ability
of GALDI MS to detect members of these classes of compounds 25 and because of the
breadth of information on the effects of genetic mutants and environmental perturbations on
these metabolites. In addition, we discuss methodological details about MS imaging of
surfaces in small-size plants in terms of sample preparation, methods for imaging, and other
experimental concerns. Finally, as a model system, we describe spatial probing of flavonoid
accumulation from light-treated stem sections.
Experimental Section
Chemicals. Standards such as long-chain fatty acids, flavonoids, and histochemical
staining reagents, diphenylboric acid 2-amino-ethyl ester (DPBA) and p-
dimethylaminocinnamaldehyde (DMACA), were purchased from Sigma-Aldrich (St. Louis,
MO). 2-Propanol-based colloidal graphite aerosol spray (Aerodag G) was obtained from
Acheson Colloids (Port Huron, MI).
Plant growth conditions. Seeds of Arabidopsis thaliana Columbia-0 ecotype (Col-0)
were surface sterilized for 1 min in 300 µl 50% ethanol, 10 min in 300 µl 50% bleach
containing 1% Tween 20, and washed with 3 changes of 300 µl sterile water. Seeds were
sown on MS media in Petri plates, and kept at 4 ˚C for 4 days. The plates were then moved to
a growth room and seedlings were transferred to pots containing soil (LC1 Sunshine Mix)
under continuous illumination (100 µmol m–2 s–1) at 23 ˚C.
Light treatment. Plants were grown in a growth room for 40 days under continuous
illumination (100 µmol m–2 s–1) at 23 ˚C in pots containing soil (LC1 Sunshine Mix). At the
time of the experiment, half of the plants (control) were left in this growth room, and the
other half were transferred into a second growth room and grown in continuous illumination
under high light intensity (750-800 µmol m–2 s–1) at 23 ˚C. Fourteen days after transfer, stem
58
samples were harvested, frozen in liquid nitrogen, and 50-80 μm thick sections were cut on a
cryostat (International Equipment Co., Needham Heights, MA).
Flavonoid staining. Cryosectioned stem samples were stained for 15-20 min with
saturated 0.25% (wt:vol) DPBA 48 with 0.02% (vol:vol) Triton X-100 and viewed with an
epifluorescent microscope (Zeiss AxioPlan II compound fluorescent microscope using an
AxioCam MRM camera (for MRC color) and AxioVision software, Carl Zeiss Inc.,
Thornwood, NY) with a FITC filter (excitation, 450-490 nm; suppression, long pass/515 nm)
according to Murphy et al. (2000) and Lazar and Goodman (2006); flavonols (kaempferol
and quercetin) stain orange, naringenin stains yellow and chlorophyll autofluorescence is red.
DMACA staining reagent (0.1% DMACA in 6N HCl:95% ethanol, 1:1) 49 was used
to detect flavonols (soluble phenols, soluble flavanols and proanthocyanidins) 50. After
staining with this reagent for 5-10 min, cryosectioned stem samples were washed three times
with distilled water and observed under the light microscope; flavonols stain blue and purple-
brown colors.
Plant sample preparation for mass spectral profiling and imaging. For leaves or
flowers, samples were attached to a stainless steel target plate of similar dimensions as a
microscopic glass slide using conductive double-sided tape (3M, St. Paul, MN). Any area
inadvertently damaged by forceps was excluded when collecting mass profiles. To attach
samples firmly onto the target plate, air pressure from a nitrogen gas cylinder was used. After
attachment, samples on the target plate were dried for 30 min under moderate vacuum (~50
Torr). In some experiments, the bottom or top halves of fresh Arabidopsis leaves were dipped
in chloroform for 30 s or 1 min to remove wax layers.
For stem sections, relocation of compounds due to water condensation is a
challenging issue because frozen sections must be thaw-mounted directly onto the stainless
steel target plate. To minimize condensation, stem sections on target slides were dried in a
semi-sealed box half-filled with desiccant by purging with a flow of dry nitrogen.
59
Colloidal graphite solution diluted 4X with 2-propanol was used for spraying
colloidal graphite onto samples. The spraying device was an airbrush, model Aztek A470
with a 0.30 mm nozzle from Testor (Rockford, IL). Spraying was at 20 psi, 8 cm from the
sample plate. Optimized spraying times as determined by mass spectral qualities were 30-40
s for leaves and 20-25 s for flowers and stem sections. Colloidal graphite dried in less than 10
s.
Sample preparation including drying process is a challenging task. We have tried
using fresh samples (without drying) for whole organ samples (leaves and whole flowers).
By comparing MS profiles from the fresh sample with those from the dried sample, we did
not observed any significant difference between them for our targeted metabolites. However,
for fresh samples, they shrunk during data collection under the sample chamber pressure
(0.17 Torr) even if they were firmly attached, and the chemical images and optical images do
not match very well. Therefore, pre-drying under the moderate vacuum pressure (~50 Torr)
was needed to prevent this. For stem cryosections, special care was needed to prevent
condensation-induced chemical relocation. Other experiments show that without such dry N2
control, water droplets formed and flavonoids were seen smeared all over the cross section.
Mass spectrometry. Mass spectra were collected with a Thermo Finnigan LTQ
linear ion trap mass spectrometer equipped with vMALDI source (Mountain View, CA). In
this mass spectrometer, nitrogen laser emission (337 nm, maximum energy of 280 µJ/pulse,
and maximum frequency of 20 Hz) was guided through a 200-µm fiber optic cable. After
passing through several optical components, 100 µm diameter laser spots were obtained at
the sample plate surface. In contrast to high vacuum sampling environments (~10-6 Torr) of
conventional MALDI TOF instruments, the LTQ mass spectrometer has the intermediate-
pressure (0.17 Torr) sampling environment which can lead to softer ionization by vibrational
and collisional cooling effects. All mass spectra were collected in the negative-ion mode and
the scanning m/z range was set from m/z 150 to m/z 1000. For unknown peaks acquired
60
directly from plant samples, peak identification were carried out by matching monoisotopic
masses of metabolites listed in databases—TAIR (http://www.arabidopsis.org/) and internal
GC-MS database—with observed mass values. Identities were confirmed by comparing
product ion spectra (up to the third generation) from unknown peaks with those from
standards or from literature data.
Mass spectral imaging and image processing of leaves. For untreated leaves, the
number of laser shots required was estimated by collecting mass spectra in a small area
(usually 5 to 10 rastering points) by turning on automatic gain control (AGC, which keeps
the ion amounts in the trap at a similar range by varying the number of laser shots) feature of
the mass spectrometer. The optimum number of laser shots was determined and fixed based
on the average number of laser shots when collecting mass spectra over the whole sample.
For chloroform-dipped leaves, however, mass spectra were collected with AGC because they
have several different surface characteristics in one sample which can lead to variations in
ion yield. The sample was scanned with a step size which ranged from 100 µm to 150 µm.
Chemically-selective images were processed by using the custom software from
Thermo (ImageQuest 1.0). Chemical abundance information was presented using either
absolute intensity values or normalized intensity values. The fractional peak intensity for the
peak of interest compared to the total ion current (TIC) of each mass spectrum was used for
normalized intensity. The mass window was from 0.8 Da to 1.0 Da.
Mass spectral imaging and image processing of flowers and stem sections. 50 µm
spacing was used for whole flower imaging and stem section imaging (oversampling).
Because of the need of complete ablation for oversampling technique 51, two methods,
selective spectra averaging and whole spectra averaging, were applied for MS imaging of
Arabidopsis flowers. Images of flowers in this article were processed with whole spectra
averaging. Experimental details of selective spectra averaging and comparison between two
methods are described in Supporting Information, and chemically-selective images from
61
Arabidopsis flower by selective spectra averaging is shown in Supporting Information Figure
S2.
Whole spectra averaging without complete ablation was designed to generate similar
MS images to those with complete ablation. Under our experimental conditions, more than
one order of magnitude decrease in normalized intensity of mass spectra was observed after 5
to 8 microscans (one microscan corresponds to getting one mass spectrum from the sample).
Therefore, a fixed number of microscans (5 for Arabidopsis flowers) was preset and data
were collected for one x,y-coordinate. All microscans were averaged and images were
generated by using the custom software from Thermo (ImageQuest 1.0).
Results and Discussion
Mass spectral profiling and imaging of metabolites from Arabidopsis leaves.
VLCFAs were readily detected as deprotonated ions ([M – H]–) from the surface of
Arabidopsis leaves (Figure 2) using GALDI in the negative-ion mode. Consistent with
previous, traditional extraction-based analyses 39, 52, 53, our analyses showed C26, C28, and
C30 fatty acids (FAs) as the most abundant such analytes on the surface of Arabidopsis
leaves (Figure 2). However, ions from flavonoid compounds were not consistently detected
from Arabidopsis leaves. After examination of the leaves used for the MS analysis, we found
that ions from flavonoids were readily detected only from the portion of leaves that were
damaged during handling when they were attached to the MS sample plate. In contrast, when
leaves were attached to the plate using air pressure to avoid damage, no ions from flavonoids
were detected (Figure 2). Because flavonoids are typically intracellular metabolites and
VLCFAs are constituents of epicuticular waxes, these results indicate that the penetration
depth of the laser is shallower than the thickness of the epicuticular wax layer of the leaf.
To confirm this observation, chemically-selective images of the partially chloroform-
dipped leaf were collected using GALDI MS. Dipping the leaf in chloroform removes the
62
surface epicuticular waxes. As shown in Figure 3, ions corresponding to VLCFAs showed
high abundance in the area of the leaf that was not dipped in chloroform, especially on the
central vein of the leaf, whereas the abundance of these ions was very low in the area of the
leaf that was dipped in chloroform. In contrast, ions specific for kaempferol (at m/z 285), and
kaempferol rhamnoside (at m/z 431) were detected mostly from the chloroform-dipped area
and from the area in which the intact leaf was grasped with forceps. Thus, minimizing
physical stress on the sample surface is critical to avoiding artificial information about the
spatial distribution of metabolites.
Intense signals at m/z 226, 210, and 194 were observed in the chloroform-dipped area.
Mass spectral profiles and chemically selective images for m/z 226 and 194 are shown in
Supporting Information Figure S1. Images for ions at m/z 226 and 194 which are 16 Da away
from ions at m/z 210 have the same specific pattern that is similar to the venation pattern of
the leaf even after chloroform dipping (see Supporting Information, Figure S1). These data
indicate that it should be possible to profile internal compounds by the chemical removal of
the surface layer from the plant sample while preserving the subsurface spatial information.
Such is not possible in conventional imaging MS. Further studies will be needed to verify
this observation, since the specific solubility of the targeted compounds and the dipping time
are expected to affect their removal and redistribution. Conceptually, depth profiling by serial
data collection can probe deeper and deeper into the tissue but additional colloidal graphite
sprayings are needed because of the decrease of signal intensities after serial collection. In
addition, ablating by laser has much smaller penetration depth than chemical extraction.
Therefore, it will take much longer time to access the same depth as that after chemical
extraction.
Mass spectral profiling of flavonoids from Arabidopsis flower. Similar to leaves,
no flavonoids or only a low apparent abundance flavonols were detected from the surface of
flower carpels or sepals. In contrast, flavonoids were easily detected from petals as highly
63
intense signals. Based on the mass spectral profile from petals and the product ion spectra for
the major peaks in this profile, at least 14 flavonoid species could be classified and identified.
These ions are listed in Table 1 with possible structural identities, and representative product
ion spectra for ions at m/z 609 and 623 are shown in Figure 4. These species basically
comprised of three different flavonols aglycones-kaempferol (K), quercetin (Q), and
isorhamnetin (I)- and their mono-, di- and triglycosides. There are some other observed ions
which correspond to other flavonol aglycones. However, in Table 1, we listed only three
flavonols and their derivatives that were thoroughly investigated by our own MSn results and
showed reproducible signals through flower by flower, or organ by organ. In GALDI MS,
flavonoids were putatively identified as follows: First, MSn experiments were performed up
to the third generation product ion spectra (n=4). Second, fragmentation patterns were
comprehensively compared with those from all available standards or with literature data.
Most product ion spectra clearly show the identity of the flavonol aglycone and the
composition of sugar moieties. In Figure 4A, for example, fragment ions at m/z 315, which
corresponds to the molecular ion of the flavonol isorhamnetin ([I - H]–), were generated after
the neutral loss (NL) of 162 Da and of 146 Da which correspond to the elimination of hexose
(Hex) residue and rhamnose (Rha) residues from parent ions at m/z 623. This series of NLs
and fragment ions show that the composition of ions at m/z 623 is I-Rha-Hex. Unlike LC-MS
and NMR techniques for flavonoid analyses 44, 45, 54-57, direct profiling by LDI MS does not
include a separation step. Therefore, there is a possibility of the presence of two or more
isobaric species at one m/z peak. In the first generation product-ion spectrum for the
precursor ions at m/z 609 (Figure 4B), the combination of NL(i), NL(iii), and the fragment
ion at m/z 285 corresponds to the composition K-Hex-Hex. In contrast, the combination of
NL(iii), NL(v), and the fragment ion at m/z 301 corresponds to the composition Q-Rha-Hex.
To overcome difficulties associated with isobaric ions, separating isobaric ions by detecting
their different specific fragment ions by tandem MS imaging may be needed 24, 25.
64
Interestingly, mass spectral profiles generated from a single petal changed
dramatically depending on the locations where the mass spectra were collected (Figure 5). By
matching structural compositions in Table 1 with m/z values of the dominant species at the
tip of the petal (Figure 5A), kaempferol and its glycosides were found to be much more
abundant over the other two flavonols and their derivatives, while the dominant species close
to the flower base (Figure 5C) are quercetin (Q), isorhamnetin(I), and their glycosides. The
spatial distribution of flavonols and their glycosides are further discussed in the following
mass spectral imaging section.
Mass spectral imaging of flavonoids from Arabidopsis flowers. As shown in
Figure 5, mass spectral profiles change dramatically within a single flower petal. To confirm
this observation that flavonoid accumulation is spatially heterogeneous in petals, chemically-
selective images for ions of flavonoid species were generated. As expected, Figure 6 shows
the difference in spatial distributions of three flavonols and their sugar moieties. It should be
noted that local distributions of flavonols (at m/z 285, 301, and 315) shown in Figure 6 are
not only from flavonol aglycones themselves but also from their glycosides because favonol
glycosides also produce ions corresponding to flavonol aglycones as fragment species by the
laser- induced fragmentation. However, the degree of this fragmentation in our sampling
environment (0.17 Torr) is usually much lower than in high vacuum environment (~10-6
Torr) 58, 59. In addition, the relative abundances of isobaric ions with different compositions at
each m/z can be estimated based on the images, although the relative abundances of isomers
with the same composition cannot be distinguished by mass spectral images. According to
Table 1, isobaric ion species may exist at m/z 447 (Q-Rha and K-Hex), 593 (K-Rha-Hex and
Q-Rha-Rha), and 609 (Q-Rha-Hex and K-Hex-Hex). By comparing locations with abundant
flavonol aglycones with those of isobaric ion species in Figure 6, higher abundances of Q-
Rha at m/z 447, K-Rha-Hex at m/z 593, and Q-Rha-Hex at m/z 609 can be expected. Among
flavonol glycosides which have the same flavonol aglycone, sub-localization of their
65
glycosides was observed in one abundant area. The ion species at m/z 755, which
corresponds to quercetin triglycosides (Q-Rha-Rha-Hex), was observed to be highly
localized around the boundary between kaempferol-localized area and quercetin-localized
area.
The heterogeneous distribution of flavonoid compounds within a single petal of an
Arabidopsis flower has never been reported or visualized in any previous studies.
Simultaneously probing metabolite location and chemical composition by using imaging MS
makes it possible to expand our understanding of biological synthetic pathways of secondary
metabolites in plants. The data presented here can be combined with a series of studies about
flower development 60, 61 and biosynthetic pathways of flavonoids 46, 47, 62, 63. In particular, the
spatial distribution of these flavonol compounds suggests that genetic regulation for
expressing enzymes involved in flavonoid biosynthesis, such as flavonoid 3′-hydroxylase
(F3′H) which converts kaempferol to quercetin by catalyzing the hydroxylation at the 3′
position in the B-ring of the flavonol 43, 64-68 and flavonol glycosyl transferase (FGT) which is
responsible for transporting sugar units to the specific hydroxyl groups of the flavonol 47, 69,
may be a time-dependant process during maturation of the Arabidopsis flower.
The ability of GALDI MS imaging and profiling shown above can also be applied to
rapid screening and metabolite profiling of mutants which are related to the biosynthesis of
flavonoids such as the transparent testa mutants.43 Combining the high spatial resolution
metabolite imaging with the mutant screening by GALDI MS has the potential to provide a
new direction to functional genomics studies.
Figure 7 shows chemically-selective images generated from the whole flower. Higher
abundances of flavonols and their glycosides on the petal surface compared to other parts of
the flower surface and consistency of the localization of flavonols through petals are clearly
depicted. In addition to flavonoids, high abundances of C29 ketone (at m/z 421) and C30 FA
(at m/z 451) were observed in a carpel and a flower pedicle, which is consistent with previous
66
studies 38-40, 53. We also found several unknown compounds including ions at m/z 978 that
were highly localized on the carpel surface. Interestingly, intense signals from ion species at
m/z 717 were observed at the stigma. This species has not been unambiguously identified,
however, its spectra aresimilar to those of condensed tannin compounds 12, 70 based on its
fragment ions (m/z 602, 339, 311, 309, 291, and 247).
It should be noted that images of those ion species that were found mainly at the
carpel and the stigma were processed with normalized intensities instead of absolute
intensities. The reason is that the average intensities from the carpel (~104) is about 10 times
lower than those from petals (~105) such that the spatial distribution at the carpel becomes
difficult to distinguish from background signals of the petal area if these ions were processed
with absolute intensities. This difference in ion yield may come from several inherent factors
such as different shape and thickness of flower parts and different surface characteristics due
to the different cuticular wax load of each flower part. Therefore, quantitative comparison
among different flower organs is limited in the direct profiling of metabolites by LDI MS.
Because total ion current is influenced directly by ion yield, normalized intensities by TIC
were considered as a reasonable processing approach in this case. In addition, different
classes of compounds are difficult to compare quantitatively even if their ions are generated
from the same sample surface due to their different ionization efficiencies. In other words,
comparing the apparent amounts of flavonoids with those of cuticular wax compounds does
not give reliable quantitative estimates.
Light–induced flavonoid accumulation in Arabidopsis stems. Flavonoid
accumulation in the stem under the high fluence of photons was chosen as a model system
for demonstrating the utility of GALDI MS imaging method. After 14 days of light treatment,
stem sections were imaged and compared to images of stem sections from control plants
(Figure 8 and 9). As shown in Figure 8, light-treated Arabidopsis stems accumulate
anthocyanin pigments and become dark purple in color (Figure 8b). Here we used two
67
histochemical staining techniques (DMACA and DPBA staining) to determine flavonoid
localization in tissue and cells in Arabidopsis stem cross-sections as revealed by bright-field
light microscopy and epi-fluorescence microscopy (Figure 8c-8f) 48, 71. DMACA is highly
sensitive and selective for flavonols50: a blue to purple-brown color (Figure 8d, shown with
black arrows) is produced. The dark-blue DMACA stain showed sieve-tube elements with
large amounts of flavanols in vacuoles (black arrows in Figure 8d). Flavonoid accumulation
was visualized in vivo by using DPBA histochemical stain via fluorescence (white arrows in
Figure 8e and 8f). Thus, although transmission or fluorescence microscopy provides high
spatial information concerning the location of ‘colored’ compounds such as flavonoids, it
lacks the ability to decipher chemical identification. On the other hand, the mass spectral
images in Figure 9 show obvious areas where kaempferol at m/z 285 and their
monoglycosides at m/z 431 preferentially accumulate in light-treated stem sections as
compared to control stem sections. Another interesting observation is that the intense signals
at m/z 749 and 765 were observed only in light-treated stem sections and their distribution is
different from that of kaempferols. Based on fragment ion patterns (data not shown), these
may be similar to substructures of the end product, proanthocyanidin (PA). This result
suggests that GALDI MS profiling and imaging can simultaneously provide spatial
distribution on plant tissue sections like microscopy and chemical identities like LC-MS.
However, the current spatial resolution (50-100 µm) is not high enough to differentiate cell
by cell in epidermis, and sampling areas with diameters in the low micron range must be
developed to achieve cell-to-cell comparison similar to light microscopy techniques.
Acknowledgement
E.S.Y. thanks the Robert Allen Wright Endowment for Excellence for support. H.I.I.
thanks the National Science Foundation for support under MCB-0416730 and MCB-0520140.
The Ames Laboratory is operated for the U.S. Department of Energy by Iowa State
68
University under Contract No. DE-AC02-07CH11358. This work was supported by the
Director of Science, Office of Basic Energy Sciences, Division of Chemical Sciences.
Appendix
Further experimental consideration about enhancing spatial resolution in mass
spectral imaging by ‘Oversampling’ was discussed in Appendix 3.
Table 1. Observed flavonoid compounds by GALDI MSn from Arabidopsis thaliana wild type (Col) flower
m/z values
[M–H]–
Observed major m/z values of
fragment ions (NLs) Compositionb Possible structural identities References
285 151, 179, 229 Kaempferol (K) Kaempferol
301 151, 179, 229, 285 Quercetin (Q) Quercetin
315 151, 179, 228, 300 Isorhamnetin (I) Isorhamnetin 72
431 285(146), K-Rha
447 285(162), 301(146) Q-Rha and K-Hex Quercetin-3-O –rhamnoside and
Kaempferol-3-O – glucoside
55
461 315(146) I-Rha
463 301(162) Q-Hex Quercetin-3-O – glucoside 56
477 315(162) I-Hex
577 285, 429, 431(146) K-Rha-Rha Kaempferol-3, 7-di-O -rhamnoside 44
593 285, 327, 429, 431(162), 447(146) K-Rha-Hex and
Q-Rha-Rha
Kaempferol-3-O -glucoside-7-O-rhamnoside
and Quercetin-di-rhamnoside
44, 54
609 285, 301, 429(180), 447(162), 463(146) Q-Rha-Hex and K-Hex-Hex Quercetin-glucoside-rhamnoside 54
623 315, 461(162), 477(146) I-Rha-Hex
739a 284, 430(309), 447, 593(146) K-Rha-Rha-Hex Kaempferol-3-O –
rhamnoside(1→2)glucoside-7-O-rhamnoside
55, 73
755 301, 447, 609(146) Q-Rha-Rha-Hex Quercetin- 3-O – rhamnoside-glucoside-7-O-
rhamnoside
57, 74
a: Fragment ions from both 1st generation and 2nd generation (739→593→) product ion spectra were included
b: K: Kaempferol, Q: Quercetin, I:Isorhamnetin, Rha: rhamnoside, and Hex: hexoside (galactose or glucose)
69
70
References
(1) Dunn, W. B.; Bailey, N. J. C.; Johnson, H. E. Analyst 2005, 130, 606-625.
(2) Sumner, L. W.; Mendes, P.; Dixon, R. A. Phytochemistry 2003, 62, 817-836.
(3) Bligny, R.; Douce, R. Current Opinion in Plant Biology 2001, 4, 191-196.
(4) Ratcliffe, R. G.; Shachar-Hill, Y. Annual Review of Plant Physiology and Plant
Molecular Biology 2001, 52, 499-526.
(5) Villas-Boas, S. G.; Mas, S.; Akesson, M.; Smedsgaard, J.; Nielsen, J. Mass
Spectrometry Reviews 2005, 24, 613-646.
(6) Fiehn, O.; Kopka, J.; Dormann, P.; Altmann, T.; Trethewey, R. N.; Willmitzer, L.
Nature Biotechnology 2000, 18, 1157-1161.
(7) Roessner, U.; Wagner, C.; Kopka, J.; Trethewey, R. N.; Willmitzer, L. Plant Journal
2000, 23, 131-142.
(8) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Science 1989,
246, 64-71.
(9) Glinski, M.; Weckwerth, W. Mass Spectrometry Reviews 2006, 25, 173-214.
(10) Sluszny, C.; Yeung, E. S.; Nikolau, B. J. Journal of the American Society for Mass
Spectrometry 2005, 16, 107-115.
(11) Zabrouskov, V.; Al-Saad, K. A.; Siems, W. F.; Hill, H. H.; Knowles, N. R. Rapid
Communications in Mass Spectrometry 2001, 15, 935-940.
(12) Ishida, Y.; Kitagawa, K.; Goto, K.; Ohtani, H. Rapid Communications in Mass
Spectrometry 2005, 19, 706-710.
(13) Fraser, P. D.; Enfissi, E. M. A.; Goodfellow, M.; Eguchi, T.; Bramley, P. M. Plant
Journal 2007, 49, 552-564.
(14) Schad, M.; Mungur, R.; Fiehn, O.; Kehr, J. Plant Methods 2005, 1, 2.
(15) Caprioli, R. M.; Farmer, T. B.; Gile, J. Analytical Chemistry 1997, 69, 4751-4760.
(16) Wu, K. J.; Odom, R. W. Analytical Chemistry 1996, 68, 873-882.
71
(17) Winograd, N. Applied Surface Science 2003, 203, 13-19.
(18) Sjovall, P.; Lausmaa, J.; Johansson, B. Analytical Chemistry 2004, 76, 4271-4278.
(19) Parry, S.; Winograd, N. Analytical Chemistry 2005, 77, 7950-7957.
(20) Eriksson, C.; Borner, K.; Nygren, H.; Ohlson, K.; Bexell, U.; Billerdahl, N.;
Johansson, M. Applied Surface Science 2006, 252, 6757-6760.
(21) Heeren, R. M. A.; McDonnell, L. A.; Amstalden, E.; Luxembourg, S. L.; Altelaar, A.
F. M.; Piersma, S. R. Applied Surface Science 2006, 252, 6827-6835.
(22) Wiseman, J. M.; Ifa, D. R.; Song, Q. Y.; Cooks, R. G. Angewandte Chemie-
International Edition 2006, 45, 7188-7192.
(23) Takats, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G. Science 2004, 306, 471-473.
(24) Cha, S. W.; Yeung, E. S. Analytical Chemistry 2007, 79, 2373-2385.
(25) Zhang, H.; Cha, S. W.; Yeung, E. S. Analytical Chemistry 2007, 79, 6575-6584.
(26) Northen, T. R.; Yanes, O.; Northen, M. T.; Marrinucci, D.; Uritboonthai, W.; Apon,
J.; Golledge, S. L.; Nordstrom, A.; Siuzdak, G. Nature 2007, 449, 1033-U1033.
(27) Mangabeira, P. A.; Gavrilov, K. L.; de Almeida, A. A. F.; Oliveira, A. H.; Severo, M.
I.; Rosa, T. S.; Silva, D. D.; Labejof, L.; Escaig, F.; Levi-Setti, R.; Mielke, M. S.;
Loustalot, F. G.; Galle, P. Applied Surface Science 2006, 252, 3488-3501.
(28) Gojon, A.; Grignon, N.; Tillard, P.; Massiot, P.; Lefebvre, F.; Thellier, M.; Ripoll, C.
Cellular and Molecular Biology 1996, 42, 351-360.
(29) Gea, L.; Jauneau, A.; Vian, B. Journal of Trace and Microprobe Techniques 1994, 12,
323-329.
(30) Jauneau, A.; Pariot, C.; Verdus, M. C.; Ripoll, C. Journal of Trace and Microprobe
Techniques 1994, 12, 51-60.
(31) Derue, C.; Gibouin, D.; Demarty, M.; Verdus, M. C.; Lefebvre, F.; Thellier, M.;
Ripoll, C. Microscopy Research and Technique 2006, 69, 53-63.
72
(32) Heard, P. J.; Feeney, K. A.; Allen, G. C.; Shewry, P. R. Plant Journal 2002, 30, 237-
245.
(33) Imai, T.; Tanabe, K.; Kato, T.; Fukushima, K. Planta 2005, 221, 549-556.
(34) Saito, K.; Mitsutani, T.; Imai, T.; Matsushita, Y.; Fukushima, K. Anal. Chem. 2008,
80, 1552-1557.
(35) Burrell, M. M.; Earnshaw, C. J.; Clench, M. R. J. Exp. Bot. 2007, 58, 757-763.
(36) Li, Y.; Shrestha, B.; Vertes, A. Analytical Chemistry 2007, 79, 523-532.
(37) Aarts, M. G. M.; Keijzer, C. J.; Stiekema, W. J.; Pereira, A. Plant Cell 1995, 7, 2115-
2127.
(38) Hannoufa, A.; McNevin, J.; Lemieux, B. Phytochemistry 1993, 33, 851-855.
(39) Jenks, M. A.; Tuttle, H. A.; Eigenbrode, S. D.; Feldmann, K. A. Plant Physiology
1995, 108, 369-377.
(40) Jenks, M. A.; Tuttle, H. A.; Feldmann, K. A. Phytochemistry 1996, 42, 29-34.
(41) Podila, G. K.; Rogers, L. M.; Kolattukudy, P. E. Plant Physiology 1993, 103, 267-272.
(42) Burchard, P.; Bilger, W.; Weissenbock, G. Plant Cell and Environment 2000, 23,
1373-1380.
(43) Shirley, B. W.; Kubasek, W. L.; Storz, G.; Bruggemann, E.; Koornneef, M.; Ausubel,
F. M.; Goodman, H. M. The Plant Journal 1995, 8, 659-671.
(44) Veit, M.; Pauli, G. F. Journal of Natural Products 1999, 62, 1301-1303.
(45) Bloor, S. J.; Abrahams, S. Phytochemistry 2002, 59, 343-346.
(46) Debeaujon, I.; Nesi, N.; Perez, P.; Devic, M.; Grandjean, O.; Caboche, M.; Lepiniec,
L. Plant Cell 2003, 15, 2514-2531.
(47) Jones, P.; Messner, B.; Nakajima, J. I.; Schaffner, A. R.; Saito, K. Journal of
Biological Chemistry 2003, 278, 43910-43918.
(48) Lazar, G.; Goodman, H. M. Proceedings of the National Academy of Sciences of the
United States of America 2006, 103, 472-476.
73
(49) Sharma, S. B.; Dixon, R. A. Plant Journal 2005, 44, 62-75.
(50) Abrahams, S.; Tanner, G. J.; Larkin, P. J.; Ashton, A. R. Plant Physiology 2002, 130,
561-576.
(51) Jurchen, J. C.; Rubakhin, S. S.; Sweedler, J. V. Journal of the American Society for
Mass Spectrometry 2005, 16, 1654-1659.
(52) Millar, A. A.; Clemens, S.; Zachgo, S.; Giblin, E. M.; Taylor, D. C.; Kunst, L. Plant
Cell 1999, 11, 825-838.
(53) Rashotte, A. M.; Jenks, M. A.; Nguyen, T. D.; Feldmann, K. A. Phytochemistry 1997,
45, 251-255.
(54) Graham, T. L. Plant Physiology and Biochemistry 1998, 36, 135-144.
(55) Kerhoas, L.; Aouak, D.; Cingoz, A.; Routaboul, J. M.; Lepiniec, L.; Einhorn, J.;
Birlirakis, N. Journal of Agricultural and Food Chemistry 2006, 54, 6603-6612.
(56) Routaboul, J. M.; Kerhoas, L.; Debeaujon, I.; Pourcel, L.; Caboche, M.; Einhorn, J.;
Lepiniec, L. Planta 2006, 224, 96-107.
(57) Tohge, T.; Nishiyama, Y.; Hirai, M. Y.; Yano, M.; Nakajima, J.; Awazuhara, M.;
Inoue, E.; Takahashi, H.; Goodenowe, D. B.; Kitayama, M.; Noji, M.; Yamazaki, M.;
Saito, K. Plant Journal 2005, 42, 218-235.
(58) Garrett, T. J.; Prieto-Conaway, M. C.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford,
G.; Yost, R. A. International Journal of Mass Spectrometry 2007, 260, 166-176.
(59) Garrett, R.; Grisham, C. Biochemistry, 2nd Edition ed., 1999.
(60) Bowman, J. L.; Smyth, D. R.; Meyerowitz, E. M. Plant Cell 1989, 1, 37-52.
(61) Smyth, D. R.; Bowman, J. L.; Meyerowitz, E. M. Plant Cell 1990, 2, 755-767.
(62) Kitamura, S.; Shikazono, N.; Tanaka, A. Plant Journal 2004, 37, 104-114.
(63) Xie, D. Y.; Sharma, S. B.; Paiva, N. L.; Ferreira, D.; Dixon, R. A. Science 2003, 299,
396-399.
(64) Boss, P. K.; Davies, C.; Robinson, S. P. Plant Physiology 1996, 111, 1059-1066.
74
(65) Burbulis, I. E.; Winkel-Shirley, B. Proceedings of the National Academy of Sciences
of the United States of America 1999, 96, 12929-12934.
(66) Holton, T. A.; Brugliera, F.; Lester, D. R.; Tanaka, Y.; Hyland, C. D.; Menting, J. G.
T.; Lu, C. Y.; Farcy, E.; Stevenson, T. W.; Cornish, E. C. Nature 1993, 366, 276-279.
(67) Landry, L. G.; Chapple, C. C. S.; Last, R. L. Plant Physiology 1995, 109, 1159-1166.
(68) Koornneef, M.; Luiten, W.; de Vlaming, P.; Schram, A. W. Arabid. Inf. Serv. 1982,
19, 113-115.
(69) Yonekura-Sakakibara, K.; Tohge, T.; Niida, R.; Saito, K. Journal of Biological
Chemistry 2007, 282, 14932-14941.
(70) Behrens, A.; Maie, N.; Knicker, H.; Kogel-Knabner, I. Phytochemistry 2003, 62,
1159-1170.
(71) Murphy, A.; Peer, W. A.; Taiz, L. Planta 2000, 211, 315-324.
(72) Wolfender, J. L.; Waridel, P.; Ndjoko, K.; Hobby, K. R.; Major, H. J.; Hostettmann,
K. Analusis 2000, 28, 895-906A.
(73) March, R. E.; Miao, X. S.; Metcalfe, C. D. Rapid Communications in Mass
Spectrometry 2004, 18, 931-934.
(74) Besseau, S.; Hoffmann, L.; Geoffroy, P.; Lapierre, C.; Pollet, B.; Legrand, M. Plant
Cell 2007, 19, 148-162.
75
Figure Captions
Figure 1. Structures of flavonol aglycones and glycosides. Arrows indicate hydroxyl
groups that are modified by glycosylation.
Figure 2. Negative-ion mode GALDI mass spectrum taken from an Arabidopsis leaf.
Mass spectra from 45 scanning points on the leaf central vein were averaged.
VLCFAs which have 24 to 30 carbons were detected. C24, C26, C28, C30
FAs correspond to tetracosanoic, hexacosanoic, octacosanoic, and
triacontanoic fatty acids.
Figure 3. An optical image of chloroform-dipped leaf (upper center) and chemically-
selective images of major compounds. Dimension of the image is 7600 µm
high × 14100 µm wide. The step size for data collection was set to 150 µm for
both x and y directions. C26 and C30 fatty acids showed high abundance on
untreated area (right column) while kaempferol and its monoglycoside
showed high abundance in forcep-grasped area and chloroform-dipped area
(left column). Ion species at m/z 210 showed very high abundances in the
dipped area (lower center). Mass spectral profile of the dipped area is shown
in Supporting Information Figure S1. All mass spectral images shown here
were presented as normalized intensities and maximum scale values for
kaempfeol, kaempferol rhamnoside, ions at m/z 210, C26 fatty acid, and C30
fatty acid were set to 2.5%, 3%, 10%, 5%, and 3% of total ion current (TIC)
value, respectively. All images were smoothed linearly.
76
Figure 4. First generation product ion spectra of ions detected at (A) m/z 623 and
(B) m/z 609 from the negative-ion mode GALDI mass spectrum of
Arabidopsis petal. The neutral loss of 146 Da (NL 146) corresponds to the
elimination of a rhamnose moiety and NL 162 corresponds to the elimination
of a hexose moiety. The loss of 163 Da corresponds to the elimination of the
radical C6H11O5°. NL 180 corresponds to subsequent water loss after NL 162.
Fragment ions at m/z 315 and m/z 285 correspond to deprotonated ions of
isorhamnetin(I) ([I – H]-) and kaempferol(K) ([K-H]-), respectively. Chemical
compositions corresponding to ions (A) m/z 623 and (B) m/z 609 are listed in
Table 1.
Figure 5. Mass spectral profiles from three locations in an Arabidopsis petal. The
optical image on the left side is the image of the whole flower which was
attached to the sample plate. The image was taken inside the mass
spectrometer before spraying colloidal graphite onto the flower. The image on
the right side is the magnified view of the petal. Dimensions are 4900 µm high
× 5200 µm wide for the whole flower image and 2822 µm high × 2445 µm
wide for the petal image. Three areas annotated with (A), (B), and (C) indicate
where the mass spectra were collected. Each area contained 50 rastering
points and the plots were the averaged mass spectra of 50 rastering points
from each area. Symbol K, Q, and I correspond to the three flavonol
aglycones listed in Table 1 and numbers above the major peaks depict
nominal m/z values listed in Table 1.
Figure 6. Chemically selective images of Arabidopsis petals. Petals in these images are
the same as in Figure 5. The step size for data collection was set to 50 µm for
77
both x and y directions. Dimension of the images is 2822 µm high × 2445 µm
wide. The nominal value in each image corresponds to the nominal m/z value
of the corresponding ions. Chemical compositions of corresponding m/z
values are listed in Table 1. All images were processed with absolute intensity
values and intensity scales were adjusted individually for the best presentation.
The images were not smoothed.
Figure 7. Chemically selective images of Arabidopsis flower. The step size for data
collection was set to 50 µm for both x and y directions. Dimension of the
images is 4740 µm high × 4600 µm wide. The nominal value in each image
corresponds to the nominal m/z value of the corresponding ions. Chemical
compositions of corresponding m/z values were listed in Table 1 except m/z
421(C29 ketone), 451(C30 fatty acid), 978(unknown), and 717(unknown).
Four mass images (421,451, 978, and 717) on the first row were processed
with normalized intensities and the maximum scale values were shown as a
percentage of the total ion current. The other images were processed with
absolute intensity values and the maximum scale value was given for each
image. All images were smoothed linearly.
Figure 8. Effect of light on flavonoid and anthocyanin content and distribution in the
inflorescence stem of Arabidopsis. Left column, grown under normal-light
regime; and right column, grown under high-light regime. (a) and (b),
Stereomicroscopic images of the inflorescence stem. Anthocyanin
accumulation in (b) is shown with stars. Bars: 500 µm. (c) and (d), Bright-
field light microscopic view of inflorescence stem cross-sections using
DMACA staining to show localization of total soluble phenols, soluble
78
flavanols and and proanthocyanidins (PAs): the black arrows show blue-
brown-purple colored sieve-tube elements of phloem cells. Bars: 20 µm. (e)
and (f), DPBA staining viewed by fluorescence microscopy: fluorescence
shows mostly selected flavonoid localization in the collenchyma under the
epidermis as patchy areas. Orange fluorescence (kaempferol and quercetin);
bright-yellow fluorescence (naringenin-chalcone); and chlorophylls red
autofluorescence (shown with white arrows). Ph, Phloem; col, collenchyma
cells; and Xy, Xylem. Bars: 40 µm.
Figure 9. Chemically selective images of Arabidopsis stems grown under normal-light
regime as control (top row) and grown under high-light regime (bottom row).
Step spacing was set to 50 µm for both x and y directions. Sizes of both stem
sections were 1100 ~ 1200 µm in diameter. All images were processed with
normalized intensities and maximum scale values were shown as a percentage
of the total ion current. The accumulation of kaemperols (m/z 285) and its
glycosides (m/z 431) in light-treated stem section is obvious compared to
control stem section. The accumulation of the unknowns at m/z 749 and 765
shows a different local distribution from kaempferols. All images were
smoothed linearly.
79
HO
OH
O
OOH
R
OH
7 3
3'
OH
HO
H
HO
H
HOHH OH
OH
OH
OH
H
OHOH
OHH HH
beta-D-glucopyranose
6-deoxy-alpha-L-mannopyranose (Rhamnose)
R: H (kaempferol)OH (quercetin)OCH3 (isorhamnetin)
Figure 1.
80
Figure 2.
81
C26
ACI
D (m
/z39
5)
C30
ACI
D (m
/z45
1)
Dip
ped
Not
dip
ped
Gra
sped
Kaem
pfer
ol(m
/z28
5)
Kaem
pfer
olRh
amno
side
(m/z
431)
m/z
210
0M
AX
Figure 3.
82
Figure 4.
83
(C)
(B)
(A)
Figure 5.
84
285
301
315
577
609
477
431
463
461
739
447
623
593
755
Figure 6.
85
m/z
285
(1.0
E6)
m/z
431
(5.0
E5)
m/z
593
(3.0
E5)
m/z
739
(5.0
E4)
m/z
301
(3.0
E5)
m/z
315
(1.5
E5)
m/z
477
(1.0
E5)
m/z
609
(1.0
E5)
m/z
421
(5.0
%)
m/z
451
(5.0
%)
m/z
978
(3.0
%)
0M
AX
m/z
755
(4.0
E4)
m/z
717
(0.5
%)
m/z
577
(1.0
E5)
Figure 7.
86
Figure 8.
87
m/z
285
(2.0
%)
m/z
431
(2.0
%)
m/z
749
(1.0
%)
m/z
765
(1.0
%)
0MA
XC
ON
TRO
L
LIG
HT
TREA
TED
Figure 9.
88
CHAPTER 4. DIRECT PROFILING AND IMAGING OF
EPICUTICULAR WAXES ON ARABIDOPSIS THALIANA BY LASER
DESORPTION/IONIZATION MASS SPECTROMETRY USING
SILVER COLLOID AS A MATRIX
Part of this chapter submitted to The Plant Journal*
Sangwon Cha, Zhihong Song, Wenxu Zhou, Basil J. Nikolau and Edward S. Yeung
Abstract
Colloidal-silver laser desorption/ionization (LDI) mass spectrometry (MS) was
employed to profile and image epicuticular wax metabolites in Arabidopsis leaves and
flowers. Mass spectral profiles of wax compounds were recorded directly from various plant
surfaces. Major wax compounds, very long chain fatty acids and their derivatives including
alkanes, primary alcohols, and ketones were successfully detected as silver adduct ions. With
the ability of small-area sampling and the excellent sensitivity, surface species of each flower
part, carpels, petals, and sepals, were profiled for the first time. In addition, mass spectral
profiles and images collected from wild-type and mutants were used for investigating
variations of wax products by normalizing the ion intensities to a reference peak, [107Ag + 109Ag]+. Differences in profiles between wild-type (Ler) and cer mutant (cer2) by LDI MS
were compared with those from GC-MS. We propose that the cer2 gene product may be
involved in either C30 fatty acid to C30 aldehyde reduction or C30 aldehyde to C29 alkane
decarbonylation in the wax biosynthesis pathway. It may also have a location specific
regulatory function.
Introduction
As a barrier to the abiotic environment, plants produce the cuticle which protects
them from external stresses, such as water, insects, and pathogens. The cuticle is composed
89
of two hydrophobic components, cutin and cuticular waxes 1, 2. A fatty acid-based polyester,
cutin serves as the structural backbone of cuticle and is covered and blended with aliphatic
long-chain fatty acids and their derivatives, cuticular wax. Functional genomics and
metabolomics studies on cuticular wax biosynthesis pathway and their transport mechanism
have been carried out mainly with Arabidopsis thaliana and achievements in this field were
thoroughly reviewed recently 3.
Several wax-deficient mutants have been screened mainly by visually examining the
phenotype, and the mutant loci have been identified for Arabidopsis thaliana, eceriferum
(cer) 4. In addition to visual screening, a condensed GC method was employed to screen the
visually identical mutant lines 5, twenty four cer loci have been identified so far. Among
CER genes, CER2 gene has been isolated and characterized by Xia et al. 6. And they
suggested that CER2 gene encodes a 47 kDa, nuclear localized protein 7. However, the
function of CER2 gene product is still unknown 3.
The wax profiles of cer mutants have been analyzed by gas chromatography mass
spectrometry (GC-MS) and, based on differences in the carbon-chain lengths and the
amounts of wax constituents, the function of cer genes and their products that are involved in
wax biosynthesis were putatively identified 5, 8-16. GC-MS methods are always concomitant
with chemical extraction of the wax compounds by organic solvents and derivatization with
volatile functional groups such as trimethylsilane groups. Therefore, these profiling
procedures are labor-intensive and the sample sizes are limited by the extraction protocol and
by the sensitivity of GC instrumentation. In addition, chemical extraction cannot be
exclusively restricted to epicuticular waxes and the degree of extraction is also ambiguous.
Recently, laser desorption/ionization mass spectrometry (LDI MS) has been
employed to analyze various wax compounds and alkane oligomers 17-20. In these studies,
silver-based matrixes yielded adduct ions for alkanes and their derivatives. The other
advantage of silver-based matrixes over other transition metal-based matrixes is that the
90
reactivity of silver to alkanes is lower than that of any other transition metals. Therefore,
silver mostly generates intact silver adduct ions without fragmentation 21. Recently, direct
LDI-MS analysis of wax compounds on Arabidopsis stems and leaves by using silver colloid
as a matrix was successfully performed 22. The advantage of colloidal silver over other silver
matrixes, such as ultrafine silver powder, silver nitrate, and AgTFA is that colloidal silver
solution can be applied more easily and more homogeneously onto hydrophobic surfaces
than other silver matrixes. Since it does not form crystals when it dries, good reproducibility
of the mass spectral signals can be expected, making it possible to examine localization of
the compounds of interest on the plant surface. In other words, colloidal silver is well suited
for imaging MS.
Here we extend the application of colloidal silver-LDI MS for the analysis of plant
cuticular wax. By rastering over the plant material, mass spectral profiles and chemically-
selective images were successfully generated for Arabidopsis leaves and flowers. After
normalization of the mass spectra with respect to the reference peak, reliable relative
abundance information was obtained that can be compared with traditional GC-MS analysis.
Compositional variations between wild-type and cer2 mutant flower parts were investigated
for the first time by silver-LDI MS and GC-MS analysis of whole-flower wax metabolites
was also performed whereas most of the previously reported metabolite profiling were
focused on leaf and stem 10, 16.
Experimental Section
Chemicals. Standards such as n-alkanes, alcohols, esters, ketones, and fatty acids,
consisting of 13–32 carbon atoms, BSTFA/TMCS (N,O-Bis(trimethylsilyl)trifluoroacetamide
with trimethylchlorosilane), and MS media (Murashige and Skoog basal salt mixture) were
obtained form Sigma-Aldrich (St. Louis, MO). A mixture of n-alkanes (C12–C60) was
purchased from Supelco (Bellefonte, PA). Water-based colloidal silver (20 ppm) was
91
purchased from Purest Colloids (Westhampton, NJ). Double-deionized water was used from
a MilliQ water purification system (Framington, MA). All other chemicals were purchased
from Fisher Scientific (Fairlawn, NJ).
Plant growth conditions. Arabidopsis thaliana ecotype Landsberg erecta (Ler-0)
and its eceriferum mutant (cer2) seeds were obtained from the ABRC. Seeds were sown on
MS media in petri dishes. The dishes were placed in the growth room for 10 days after
keeping at 4oC for 4 days. On 15th day the plants were transferred in soil for continuous
growth. Leaves were collected on 26th days, and flowers were collected on 34th days.
Growth room was at 24°C with an illumination at 85 μE m-2s-1 and 100% relative humidity.
Plant sample preparation for LDI mass spectral profiling and imaging. After
collecting samples from plants, plant materials were attached onto a stainless steel target
plate of similar dimensions as a microscope glass slide using a conductive double-sided tape
(3M, St. Paul, MN). To avoid damage to the plant sample surface, contact area by forceps
was minimized and air pressure from a nitrogen gas cylinder was used for attaching samples
firmly onto the sample plate. All samples were dried in moderate vacuum (approximately
~50 Torr) for 30–60 min. As shown in Figure 1, carpels, petals, and sepals were separated
and attached individually for metabolite profiling of flower parts. Three replicates of leaves
and whole flowers, and five replicates of flower parts for each genetic type were prepared.
The spraying device for colloidal silver solution was composed of a computer-
controlled syringe pump with a 500 μL syringe (Kloehm LTD., Las Vegas, NV), MicroFlow
PFA-ST nebulizer (Elemental Scientific Inc., Omaha, NE) with a 0.25 mm i.d. sample uptake
capillary, and a helium gas cylinder. Spraying time and speed was programmed by the
software “WinPump®” which is provided by Kloehm. To obtain a homogeneous coverage of
silver particles, several intervals of spraying with a small volume of colloidal silver solution
were more desirable than one-time spraying with a large volume because silver particles tend
to aggregate when they are dried. Therefore, the parameters for spraying were optimized as
92
100 μL/min flow rate with 12.5 μL of spraying volume, and 8 sets of spraying and air-drying
were performed for each plant sample. The distance between the nebulizer tip and the sample
was 12.4 cm and the area covered by colloidal silver on the sample plate was 4.5 cm2.
Spraying onto the leaf was performed without moving the sample plate because leaf sizes are
much smaller than the spray area.
Plant sample preparation for gas chromatography-mass spectrometry. Twenty
five flowers or whole leaves of a plant were collected and weighed. An aliquot of internal
standard hexadecane was added to the surface of the plant material. The plant material was
completely immersed in chloroform for 60 s. The cuticular wax was extracted and dried
under nitrogen gas. Samples were derivatized using BSTFA/TCMS (65˚C, 30 min) for
GC/MS analysis. 3 replicates of leaves and flowers were prepared.
Mass spectrometry. For LDI mass spectral profiling and imaging, a Thermo
Finnigan LTQ linear ion trap mass spectrometer equipped with vMALDI source (Mountain
View, CA) was used. Fiber-optic guided nitrogen laser (337 nm, maximum energy of 280
µJ/pulse, and maximum frequency of 20 Hz) was used as a laser source. The laser spot size
was 100 µm at the target plate surface. The intermediate-pressure (0.17 Torr) sampling
environment was kept by nitrogen gas flow control and this allows softer ionization
compared to high vacuum environments (~10-6 Torr). All mass spectra were collected in the
positive-ion mode and the scanning range was set to m/z 200 to m/z 1000.
For unknown peaks acquired directly from plant samples, peak identification was
carried out by matching monoisotopic masses with those detected from GC-MS experiment.
For overlapped peaks due to two stable silver isotopes and similar monoisotopic masses, the
major compounds detected were determined by comparing the first generation product ion
spectra from unknown peaks with those from standards or from literature data.
GC-MS analysis was performed with an Agilent 6890 GC interfaced to a 5973 mass
spectrometer (Agilent Technologies). A HP-5ms column (30 m x 0.25 mm i.d. coated with a
93
0.25 µm film, Agilent Technologies) was used, and temperature gradient was programmed
from 80 to 320 °C at 5°C/min with He flow rate at 2.2 mL/min. Operating parameters were
set to 70 eV (electron ionization) for ionization voltage and 280 °C for interface temperature.
The GC-MS data files were deconvoluted by NIST AMDIS software, and searched against
our laboratory’s standard compounds library and the NIST compounds library.
LDI Mass spectra collection for profiling and imaging. For leaves and flower parts,
the number of laser shots and the laser intensity scale were examined by collecting mass
spectra in a small area (usually 5 to 10 rastering points) by turning on the automatic gain
control feature (AGC, which keeps the ion amounts in the trap at a similar range by varying
the number of laser shots) of the mass spectrometer. After optimizing the laser parameters, a
fixed number of laser shots without AGC was used for collecting mass spectra over the
whole sample. For whole flower imaging, however, varying numbers of laser shots that were
controlled by AGC were applied because each flower part has different surface
characteristics and wax loads which can affect ion yields. The sample was scanned with a
step size which ranged from 50 µm to 100 µm.
Mass spectral images for whole flowers were processed by using the custom software
from Thermo (ImageQuest 1.0). The normalized intensity value which is defined as the
fractional peak intensity compared to the total ion current (TIC) of each mass spectrum was
used for presentation of chemical abundance information. The mass window was set to 0.8
Da – 1.0 Da. For comparing relative abundances among samples, intensities relative to the
peak intensity at m/z 215.8 ([107Ag + 109Ag]+) were used.
Results and Discussion
Silver-LDI mass spectral profiling of metabolites from Arabidopsis leaves. Figure
2 shows the averaged silver-LDI mass spectrum taken from an Arabidopsis thaliana wild
type (Ler) leaf surface. As shown in the inset of Figure 2, the Ag2+ ion at m/z 215.8
94
corresponding to [107Ag + 109Ag]+ showed the highest intensity, and this was consistent
through all mass spectra collected under our experimental conditions. Therefore, the relative
intensity with respect to the peak intensity at m/z 215.8 was used as the intensity scale for
quantification purposes in interpreting the mass spectra of wax compounds.
Because two stable isotopes of silver are present with similar abundances (107Ag:
51.839 % and 109Ag: 48.161 %), each hydrocarbon compound produces a quartet of silver
adduct peaks. This quartet pattern helped to distinguish sample peaks from noise, but the
multiple peaks induced spectral complexity due to the overlap of compounds different by 2
Da in molecular weight. As shown in Figure 2, peaks with odd nominal masses were the
major peaks for the cuticular wax compounds because intensities of the two other isotopic
peaks (due to either 13C or 2H) were much smaller in the mass range where cuticular wax
compounds were detected. Therefore, only peaks with odd nominal masses will be discussed
in this article to simplify the interpretation of the mass spectra. From the mass spectra
collected from Arabidopsis plants, it was hard to identify individual quartet peaks directly
due to overlap, but the m/z ranges which have signals from cuticular wax compounds were
easily grouped as shown in Table 1.
In Table 1, it is obvious that wax load amounts from parallel GC-MS analysis and the
sum of relative intensities from silver-LDI mass spectra according to m/z ranges have a
similar tendency but are not in direct proportion. There are several reasons for this
phenomenon. One of the main reason is the sampling depth by LDI is shallower than that by
the current chemical extraction protocol for GC-MS analysis. Therefore, the silver-LDI mass
spectral profile directly from cuticular wax surfaces does not represent all the cuticular wax
compounds. The shallower sampling depth by LDI may bear the possibility of serial profiling
of cuticular wax compounds according to the depth of the wax layer. For example, plant
sterols and triterpenoids were extracted and constituted about 6 % (~1.5 µmol/g per fresh
weight) of the total wax load. With colloidal silver, plant sterol standards were detected with
95
a very low detection limit, ~200 amol/10000 µm2 (sampling area of the laser). However,
there was no sterol species detected in the LDI mass profiles obtained directly from leaf
surfaces. This indicates that there may be no plant sterols at the outermost epicuticular wax
surface. In addition, as previously reported 22, there are differences in ionization efficiencies
of the wax compounds depending on their functional groups. Therefore, it is difficult to relate
observed intensities directly to actual amounts of wax compounds. In silver-LDI MS, wax
standards which have 18 - 32 carbons showed good peak reproducibility in terms of
intensities but compounds which have smaller carbon number, such as C16 fatty acid, showed
relative intensity variations depending on laser intensities and concentrations. Similar to the
previous report 22, alkanes showed the smallest ionization yields and fatty acids showed the
highest ionization yields for the same molar amount. This observation seems to be consistent
in the wax compound profile. In Table 1, the relative amounts were higher in LDI than those
in GC for m/z regions which contain fatty acid species such as m/z 475 - 477, 503 - 505 and
557 - 561, and the relative abundances were much lower in LDI than those in GC for m/z
regions where major alkanes are present, such as m/z 515-519 and 543-547. Finally, the
relative extraction efficiencies for the different classes of compounds in the GC-MS protocol
are not constant.
Despite overlaps due to silver isotopes and similar molecular weights, many of the
wax compounds in Table 1 can be recognized independently by focusing on one of two silver
isotope adduct ion peaks. In the m/z range of 571-575, for example, C33 alkane and C32
alcohol ion species overlapped each other at m/z 573 as [C33 alkane + 109Ag]+ and [C32
alcohol + 107Ag]+, but each was uniquely detected at m/z 571 as [C33 alkane + 107Ag]+ and at
m/z 575 as [C32 alcohol + 109Ag]+, respectively. However, isobaric ions such as C29 alkane
and C28 aldehyde cannot be distinguished under the current experimental conditions and the
major compounds found in LDI MS among isobaric ions need to be identified putatively by
examining fragmentation patterns in their first generation product ion mass spectra. There,
96
for ion species which contain long chain fatty acids such as m/z 503, 559 and 587 or 589,
losses of water (18 Da) and formic acid (46 Da) were always observed predominantly with
very similar ratios to those observed from fatty acid standards. This may be because of the
high ionization efficiency of fatty acids, but it may also suggest that fatty acids are localized
at the outer cuticular wax layer. For the ion species at m/z 515 and 543 which contain major
alkanes, the fragmentation patterns matched well with those of alkane standards. In spite of
the low ionization efficiency of alkanes, they can be easily detected because the molar
amounts of alkanes in the GC-MS analysis are about 20 to 30 fold higher than isobaric
aldehydes. However, the inability to detect aldehydes separately under the current
experimental conditions (because the reduction steps from fatty acids to aldehydes cannot be
traced by this method) is the main limitation of the silver-LDI MS method. Accurate mass
measurement with high mass resolution could be the solution for this problem. Also, as
indicated in Table 1, unidentified contaminants showed peaks at m/z 533-537. Therefore,
abundance information of C28 fatty acid could not be extracted unambiguously from silver-
LDI mass spectra.
Finally, there were several unknown wax compounds detected in m/z 585 – 673. Two
groups of unknown ion species were found in this m/z range (annotated as ‘Unk-A’ and
‘Unk-B’ in Table 1). Unk-A was detected at m/z 585, 613, 641, and 669 as 107Ag adduct ions
and Unk-B was detected at m/z 597, 625, and 653. Ions in each group are 28 Da away from
each other and this difference corresponds to a C2H4 group. For Unk-B, losses of multiples of
42 Da, which could correspond to propene, were observed and either single loss or quadruple
loss of 42 Da was found as a major fragment ion. For Unk-A, the product ion spectra were
very complicated due to overlap with minor unknown ions.
Comparison of leaf cuticular waxes between wild-type (Ler) and cer2 mutant.
Based on parallel GC-MS cuticular wax results and putative identification of the major
contributing compounds within overlapped isobaric species by tandem MS experiments, wax
97
metabolites that were detected with confidence were summarized in Table 2. For wild-type
(Ler) and cer2 mutant, the relative intensities by silver-LDI MS and the amounts quantified
by GC-MS for corresponding wax metabolites were also shown in Table 2. Ratios of the wax
metabolites were calculated and presented as log2(cer2/Ler) values in Figure 3. From Table 2
and Figure 3, several observations could be summarized. First, all alkanes were detected with
significantly lower intensities or lower amounts from cer2 mutant leaves by both methods.
Second, differences in the amounts of primary alcohols between wild-type and cer2 mutant
were less significant than among other classes of compounds in both methods. Third, from
the silver-LDI MS results, C24 and C26 fatty acids showed significantly smaller intensities in
cer2 than in wild-type, but longer chain length fatty acids, C30 and C32 fatty acids, were
found as more abundant in cer2 than wild-type leaves. In GC-MS analysis, however, all fatty
acids showed a small increase in cer2 mutant. This may also suggest the possibility of
differential integration of fatty acids according to the depth of the cuticular wax layer.
Discussions about the function of the cer2 gene product are given in the last section by
combining the observations above with the results from flower parts.
Silver-LDI mass spectral imaging of Arabidopsis whole flower. For
demonstrating mass spectral imaging as a rapid metabolite profiling tool, a flower was
chosen among Arabidopsis plant organs because flowers are composed of several sub-parts
that have small dimensions, such as carpel, petal, sepal, and stamen. Mass spectral images of
wild-type (Ler) flowers are shown in Figure 4. Because different flower organs have different
surface characteristics and wax loads, the total ion current (TIC), which is directly affected
by ion yields, varies among the flower organs. Therefore, all chemically-selective images for
whole flowers were processed as a normalized intensity defined as the fractional peak
intensity compared to the TIC of each mass spectrum. In Figure 4, the ion species at m/z 529,
which mainly corresponds to [C29 ketone + 107Ag]+, was found as highly localized at the
flower stem and the carpel. Major alkane species, such as [C29 alkane + 107Ag]+ and [C31
98
alkane + 107Ag]+ at m/z 515 and 543, respectively, were found at all three organs as the
major species but they showed relatively low abundances in the tips of petals. Two major
wax compounds, C31 alkane and C30 alcohol, contribute to mass signals in the m/z range 543
– 547. Both wax compounds contributed the intensity at m/z 545 as [C31 alkane + 109Ag]+ and
[C30 alcohol + 107Ag]+. This overlapped image was clearly separated into two images by
choosing different silver isotope ions, i.e. [C31 alkane + 107Ag]+ at m/z 543 and [C30 alcohol + 109Ag]+ at m/z 547, as shown in Figure 4. From the images for m/z 543 and 547, it is clear that
the intensity at the petals and at the center part of the sepals for m/z 545 is mainly from C31
alkane, not C30 alcohol. At the tips of petals, fatty acids ([C18:3 fatty acid + 107Ag]+ and [C26
fatty acid + 107Ag]+ at m/z 385 and 503, respectively) were found at higher amounts than
other wax compounds.
For comparing wax compositions between wild-type (Ler) and cer2 mutant, serial
mass spectral images were generated with a 0.8 Da mass window and a 2.0 Da step size for
the m/z ranges where wax compounds were detected, and chemically selective images for
four m/z values that showed distinct differences between wild-type (Ler) and cer2 mutant
were shown in Figure 5(A). In figure 5(A), it is obvious that C29 alkane and C29 ketone ([C29
alkane + 107Ag]+ and [C29 ketone + 107Ag]+ at m/z 515 and 529, respectively) showed much
lower abundances in cer2 mutant than in wild-type. In contrast, C30 and C32 fatty acids ([C30
fatty acid + 107Ag]+ and [C32 fatty acid + 109Ag]+ at m/z 559 and 589, respectively) showed
higher abundances in cer2 mutant than in wild-type. Interestingly, as clearly visualized in
Figure 5(B), both fatty acids showed high abundances at the tip of the petal, while in the
carpel only C30 fatty acid was found in high abundance.
Silver-LDI mass spectral profiling of Arabidopsis flower parts. Imaging whole
flowers can provide information about wax load variations between wild-type and mutants
with localization information in a very short time, but there are also limitations. Even if
whole flowers are flattened and attached onto the mass spectrometer sample plate with extra
99
care, overlaps among flower organs cannot be avoided. In addition, the degree of wax
variation between wild-type and mutants was hard to recognize accurately by processing
images with normalized intensities. Therefore, profiling wax compounds with individual
flower parts and comparing their mass profiles based on relative intensities to the reference
peak at m/z 215.8 ([107Ag + 109Ag]+) were more desirable. Three major parts of Arabidopsis
flowers, carpels, petals, and sepals were prepared separately and mass spectra were collected
serially from these parts with a 50 µm step movement. The averaged mass spectra for three
flower parts of wild-type (Ler) and its cer2 mutant are shown in Figure 6 and more detailed
peak identifications with abundances for these flower parts are listed in Table 3. In contrast
to leaf wax profiling, peak identification was more straightforward because fewer silver
adduct ion peaks were observed per flower part. Also, the major silver adduct ion peak
pattern was a quartet which corresponds to only one wax compound. As discussed previously,
the relative intensities for each class of compounds are different. In particular, intensities
corresponding to fatty acids were significantly higher. Therefore, to assess the total wax
loads between wild-type and mutants, comparing the sum of intensities from C27 or higher
alkanes, that are known to be the most abundant wax metabolites 3, is more appropriate than
comparing the total intensities from all classes of wax metabolites. As shown in Table 3, the
sum of intensities from C27, C29, and C31 alkanes was higher in wild-type than in cer2 mutant
throughout all three flower parts. Therefore, it is obvious that a global reduction of wax loads
in cer2 mutant exists. However, for carpels and petals, C31 alkane showed a higher
abundance in cer2 mutant than in wild-type.
In the carpel (Figure 6(A) and Table 3(A)), C29 alkane at m/z 515 and C29 ketone at
m/z 529 were detected as the major wax metabolites in wild-type (Ler), as expected from the
whole flower imaging results. However, in cer2 mutant, the relative intensities for these
metabolites were lower by about 5.5 fold. In contrast, the relative intensity for C30 fatty acid
at m/z 559 was higher by about 4.0 fold in cer2 mutant. In the petal (Figure 6(B) and Table
100
3(B)), peaks corresponding to C18:3, C24, C26 fatty acids (at m/z 385, 475, and 503,
respectively) and C29 alkane at m/z 515 showed high abundances in wild-type (Ler). However,
in cer2 mutant, the relative intensities for all major peaks in wild-type (Ler) were lower by
more than 2.6 fold. In contrast, the longer fatty acids, C30 and C32 at m/z 559 and 589,
showed about 4.8 fold higher intensities for cer2 mutant. In the sepal, (Figure 6(C) and Table
3(C)), C29 and C31 alkanes and C28 and C30 alcohol were detected as major peaks in wild-type
(Ler) and the wax metabolite profile was similar to that from the leaf. In the cer2 sepal, the
major differences from the wild-type sepal were lower intensities for C29 alkane (~ 6.5 fold),
C29 ketone (~ 3.5 fold), and C26 fatty acid (~ 5.7 fold). However, C30 fatty acid and alcohol
showed increases in the cer2 sepal. Interestingly, unknown wax metabolites were detected
for both wild-type and mutant at the sepal and showed lower abundances in the cer2 mutant.
GC-MS wax metabolite profiling for the whole flower were performed and results are shown
in Figure 7. The major difference from GC-MS wax metabolite profile for leaves was that no
aldehydes were detected in whole flower analysis.
It should be noted that at least 75 whole flowers (25 flowers per sample and 3
replicates) were needed to get reproducible wax metabolite profiles from GC-MS, whereas,
in silver-LDI MS profiling, only 5 flowers were used for collecting the wax metabolite
profiles of each kind of flower part.
To compare the variations of each wax metabolite, ratios of metabolites in wild-
type(Ler) and cer2 mutant flower parts were presented in Figure 8. Several features from
silver-LDI MS could be summarized and compared with GC-MS results. First, C30 and C32
fatty acids showed higher abundances in cer2 mutant than in wild-type for all flower parts.
The increase in C30 fatty acid was most pronounced in the petal in terms of both intensities
and ratios. This increase was also observed in whole-flower analysis by GC-MS. However,
similar to the results from leaf analysis, C24 and C26 fatty acids were lower in amounts from
silver-LDI MS for all cer2 mutant flower parts whereas they were at about the same level or
101
even higher amounts from GC-MS of cer2 mutant flowers. Second, C29 ketone was obviously
lower in all cer2 mutant flower parts and the decrease was the largest at the carpel in terms of
both intensities and ratios. The decrease in C29 ketone was also seen in GC-MS results. Third,
for primary alcohols, their changes were less obvious than the other classes of wax
compounds for flower parts. This was also consistent with GC-MS results. Fourth, C27 and
C29 alkanes were significantly lower in cer2 mutant flowers. However, C31 alkane showed
the smallest decrease in ratio at cer2 sepals and even increased in cer2 carpels and petals. In
GC-MS results, C31 alkane showed the smallest change in ratio among alkane species. It
should be emphasized that GC-MS results are from whole flower waxes and not from a
single organ. Therefore, differential wax load changes from flower part to flower part cannot
be recognized by GC-MS experiments.
Prediction of regulation of wax synthesis by cer2 gene product. By combining the
observations from leaves and flower parts with known cuticular wax biosynthetic pathways
for Arabidopsis thaliana, we propose that cer2 mutation may be involved either in the C30
fatty acid to C30 aldehyde reduction pathway, as had previously suggested for cer13 by
Rashotte et al. 16, or in the C30 aldehyde to C29 alkane decarbonylation pathway. Because
aldehydes were not readily detected from flower parts, deciding between the two pathways
mentioned above was not possible based only on silver LDI MS profiling. Similar to wax
load changes of cer13 mutant stems (16, levels of C29 products, C29 alkane and C29 ketone,
were lower and levels of C30 primary alcohol are higher in cer2 mutant for all flower parts
and leaves. This is also consistent with GC-MS results. . In addition, accumulation of C32
fatty acid was observed in petals which may be caused by further elongation of the
accumulated C30 fatty acid. However, increases of C31 alkane amounts were only observed in
carpels and petals but not in sepals and leaves. The degrees of increases for C30 fatty acid in
cer2 mutants were much larger in carpels (~4 fold) and petals (~4.8 fold) than in leaves (~1.2
fold) and sepals (~1.4 fold), so lower amounts of subsequent elongation from C30 to C32 fatty
102
acid can be expected in sepals and leaves. Therefore, the relatively smaller amounts of C30
fatty acid in cer2 sepals and leaves may not compensate for the general wax load reduction in
cer2 mutants and cause decreases in C31 alkanes in cer2 sepals and leaves. However, there is
also the possibility of differential regulation of elongation from C30 to C32 fatty acid by cer2
gene product according to flower parts and leaves that is similar to differential regulation
between the stem and the leaf 10.
Conclusions
Wax metabolite profiling by LDI MS with colloidal silver as a matrix has several
advantages over GC- MS approaches. First, sample preparation is much simpler because it
does not require any extraction and derivatization steps. Second, the amounts of samples
required are about 20 to 25 times smaller than those in GC-MS. Third, localization
information of cuticular waxes can be preserved because of the direct sampling capability.
With the use of homogeneous colloidal silver coating on plant surfaces, wax metabolite
distributions on each flower part was successfully profiled for the first time. However, there
are still challenging issues in this method. Overlap due to isobaric ions, especially alkanes
and aldehydes, needs to be resolved to provide more accurate measurements of changes in
wax profiles. It may be solved by employing exact mass measurements with high mass
resolution. In addition, different ionization efficiencies for different class compounds make it
difficult to compare actual abundances between two different classes of wax metabolites
directly from the mass spectra. This may be resolved by systematic studies on response
factors as a function of metabolite classes and sizes. Last, because a mass spectral database
of metabolites for silver LDI-MS does not exist currently, identification of wax compounds is
less certain than GC-MS.
By performing reproducible spraying of colloidal silver solution onto the plant
sample surface by the finely-controllable spraying device, homogeneous and constant surface
103
coverage of colloidal silver particles was achieved. This made it possible to use the intensity
of silver dimer ions as a reference to normalize all mass spectra. In this way, comparison of
the relative wax abundances between wild-type and mutants was successfully made. Wild-
type/mutant wax profile comparisons based on silver-LDI MS results generally showed good
agreement with those based on GC-MS results. However, major variance was present
between the two methods, especially for C26 and smaller fatty acids. It may be because of
sampling depth differences and further investigation is needed to identify this bias.
Lastly, a data processing scheme for the rapid comparison between wild-type and
mutant can be proposed as shown in Figure 9. The basic idea involves two general rules. First,
the two largest peaks in a quartet pattern have about the same intensities because of the
similar natural abundances of the two silver isotopes. Second, changes in the two adjacent
peaks should be similar if only one metabolite is associated with this change. Figure 9 shows
the potential of the silver-LDI MS method as a high-throughput mutant screening tool. Future
work would involve verification of this scheme with a larger data set.
Acknowledgement
E.S.Y. thanks the Robert Allen Wright Endowment for Excellence for support. The
Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University
under Contract No. DE-AC02-07CH11358. This work was supported by the Director of
Science, Office of Basic Energy Sciences, Division of Chemical Sciences.
104
Table 1. Possible wax compounds in Arabidopsis thaliana wild type (Ler) leaves according
to mass-to-charge ratio ranges where wax compounds were detected by silver-LDI MS
m/z range Possible ion species a Wax loadc (nmol/g fresh weight)
LDI Relative Intensity (%)e
361-365 C16:1 fatty acid (361) C16:0 fatty acid (363) Unknown contaminant (363)
245.83 ± 39.88 455.39 ± 108.81
dn.a. f4.45 ± 0.53
385-393 C18:3 fatty acid (385) C18:2 fatty acid (387) C18:1 fatty acid (389) C18:0 fatty acid (391)
dn.d. 87.74 ± 18.10
571.67 ± 102.37 119.34 ± 41.54
14.18 ± 0.36
475-477 C24 fatty acid (475) 28.74 ± 3.64 9.85 ± 1.09
487-491 C27 alkane (487) C26 aldehyde (487) C26 alcohol (489)
1141.48 ± 46.00 33.13 ± 2.37
171.47 ± 27.50 15.63 ± 1.41
503-505 C26 fatty acid (503) C27 iso-alcohol (503)
92.45 ± 19.20 18.88 ± 2.50 20.33 ± 2.19
515-519 C29 alkane (515) C28 aldehyde (515) C28 alcohol (517)
8242.73 ± 148.85 403.24 ± 13.63 412.27 ± 47.53
35.97 ± 1.37
529-535 C29 ketone (529) C28 fatty Acid (531) C29 iso-alcohol (531) Unknown contaminant (533)
19.95 ± 1.69 260.95 ± 18.01
11.53 ± 0.58 n.a.
f8.38 ± 0.18
543-547 C31 alkane (543) C30 aldehyde (543) C30 alcohol (545)
9014.37 ± 378.82 313.13 ± 21.62 118.37 ± 22.01
32.79 ± 1.56
559-561 C30 fatty acid (559) C31 iso-alcohol (559)
12.41 ± 2.52 17.76 ± 1.48 6.07 ± 0.29
571-575 C33 alkane (571) C32 aldehyde (571) C32 alcohol (573)
3420.03 ± 170.57 132.80 ± 13.35
59.83 ± 6.86 20.50 ± 1.33
585-589 Unk-A1(585) C32 fatty acid (587) 6.63 ± 1.02 6.28 ± 0.23
597-603 Unk-B1(597)b C35 alkane (599) C34 alcohol (601)
n.a. 123.05 ± 17.23
3.79 ± 0.43 14.13 ± 0.77
613-617 Unk-A2 (613) C34 fatty acid (615)
n.a. 4.38 ± 0.41 4.56 ± 0.29
625-629 Unk-B2 (625) n.a. 12.00 ± 1.14
641-645 Unk-A3 (641) n.a. 5.74 ± 0.01
653-657 Unk-B3 (653) n.a. 6.14 ± 0.47
669-673 Unk-A4 (669) n.a. 4.70 ± 0.87
a. Cuticular wax compounds (nominal m/z values of 107Ag adduct ions) listed are from identified compounds in parallel GC-MS analysis.
b. Unk: Unknown ion species. Unknown species were grouped by fragmentation patterns in their first generation product ion spectra.
c. Wax load values are from GC-MS for three replicates with ± standard error d. n.a. : not applicable, n.d.: not detected e. Relative intensity (%) is defined as the fractional intensity of the intensity at m/z 215.8 which corresponds to [107Ag +
109Ag]+, and the values are listed with ± standard error. f. Due to the overlap with unknown contaminants, only the relative intensity at m/z 361 was used for m/z 361-365 and
only relative intensities at m/z 529 and 531 were included for m/z 529-535.
105
Table 2. Major cuticular wax compounds with their abundances in Arabidopsis thaliana
leaves of wild-type (Ler) and cer2 mutant
Class [M + Ag]+ Major ion species Silver LDI (R.I., %)a GC (nmol/g fresh weight)b
Ler cer2 Ler cer2
Alkane C27 487.3 [C27 alkane + 107Ag]+ 6.38 ± 0.69 2.17 ± 0.23 1141.48 ± 46.00 364.72 ± 23.42
C29 515.4 [C29 alkane + 107Ag]+ 10.63 ± 0.39 3.48 ± 0.33 8242.73 ± 148.85 1011.67 ± 53.85
C31 543.4 [C31 alkane + 107Ag]+ 12.65 ± 0.99 7.00 ± 0.28 9014.37 ± 378.82 2202.02 ± 95.96
C33 571.4 [C33 alkane + 107Ag]+ 7.37 ± 0.66 3.71 ± 0.12 3420.03 ± 170.57 132.70 ± 8.32
C35 599.4 [C35 alkane + 107Ag]+ 5.47 ± 0.51 3.33 ± 0.02 123.05 ± 17.23 11.10 ± 0.59
Alcohol C26 491.3 [C26 alcohol + 109Ag]+ 2.63 ± 0.43 2.97 ± 0.60 171.47 ± 27.50 242.22 ± 20.44
C28 519.4 [C28 alcohol + 109Ag]+ 9.01 ± 1.08 7.24 ± 0.64 412.27 ± 47.53 268.81 ± 46.16
C30 547.4 [C30 alcohol + 109Ag]+ 5.25 ± 0.33 9.04 ± 0.95 118.37 ± 22.01 132.73 ± 20.22
C32 575.4 [C32 alcohol + 109Ag]+ 4.01 ± 0.49 5.31 ± 0.44 59.83 ± 6.86 15.69 ± 1.72
C34 603.4 [C34 alcohol + 109Ag]+ 1.26 ± 0.14 2.39 ± 0.09 3.79 ± 0.43 5.33 ± 0.66
Fatty acid C24 475.3 [C24 fatty acid + 107Ag]+ 5.35 ± 0.80 2.74 ± 0.16 28.74 ± 3.64 48.11 ± 6.96
C26 503.3 [C26 fatty acid + 107Ag]+ 10.95 ± 1.62 3.48 ± 0.33 92.45 ± 19.20 223.87 ± 20.64
C30 559.4 [C30 fatty acid + 107Ag]+ 3.71 ± 0.23 4.67 ± 0.10 12.41 ± 2.52 18.80 ± 2.07
C32 589.4 [C32 fatty acid + 109Ag]+ 1.16 ± 0.13 2.50 ± 0.10 6.63 ± 1.02 11.04 ± 1.50
Ketone C29 529.4 [C29 ketone + 107Ag]+ 2.64 ± 0.13 1.42 ± 0.07 19.95 ± 1.69 20.12 ± 2.01
a. Relative intensity (R.I.) (%) is defined as the fractional intensity of the intensity at m/z 215.8 which corresponds to [107Ag + 109Ag]+ and the values are listed with ± standard error for three replicates.
b. Wax load values are from GC-MS with ± standard error for three replicates
106
Table 3. Major cuticular wax metabolites detected for each flower part by silver-LDI MS for
Arabidopsis thaliana wild-type (Ler) and cer2 mutant.
Table 3(A) Carpel
Class [M + Ag]+ ion species R. I.(Ler) (%) R. I.(cer2) (%) Log2(cer2/Ler)
Alkane C29 515.4 [C29 alkane + 107Ag]+ 16.20 ± 0.31 2.93 ± 0.18 -2.47 ± 0.08
C31 543.4 [C31 alkane + 107Ag]+ 3.64 ± 0.09 5.84 ± 0.84 0.68 ± 0.18
Alcohol C28 519.4 [C28 alcohol + 109Ag]+ 3.60 ± 0.17 2.03 ± 0.05 -0.83 ± 0.06
C30 547.4 [C30 alcohol + 109Ag]+ 2.85 ± 0.17 3.04 ± 0.43 0.09 ± 0.19
C32 575.4 [C32 alcohol + 109Ag]+ 2.52 ± 0.36 2.19 ± 0.82 -0.20 ± 0.48
Fatty acid C18:3 385.3 [C18:3 fatty acid + 107Ag]+ 3.24 ± 0.18 3.01 ± 0.70 -0.11 ± 0.29
C24 475.3 [C24 fatty acid + 107Ag]+ 2.60 ± 0.09 1.24 ± 0.12 -1.06 ± 0.12
C30 559.4 [C30 fatty acid + 107Ag]+ 1.86 ± 0.19 7.37 ± 1.09 1.99 ± 0.22
Ketone C29 529.4 [C29 ketone + 107Ag]+ 21.48 ± 0.98 4.22 ± 0.67 -2.35 ± 0.20
107
Table 3(B) Petal
Class [M + Ag]+ ion species R. I.(Ler) (%) R. I.(cer2) (%) Log2(cer2/Ler)
Alkane C27 487.3 [C27 alkane + 107Ag]+ 2.91 ± 0.20 2.07 ± 0.14 -0.48 ± 0.12
C29 515.4 [C29 alkane + 107Ag]+ 15.87 ± 1.03 6.10 ± 0.89 -1.38 ± 0.18
C31 543.4 [C31 alkane + 107Ag]+ 6.63 ± 0.37 8.97 ± 0.73 0.44 ± 0.12
Alcohol C26 491.3 [C26 alcohol + 109Ag]+ 2.24 ± 0.06 3.07 ± 0.40 0.45 ± 0.16
C28 519.4 [C28 alcohol + 109Ag]+ 4.88 ± 0.17 3.48 ± 0.26 -0.49 ± 0.10
C30 547.4 [C30 alcohol + 109Ag]+ 2.56 ± 0.39 3.32 ± 0.13 0.38 ± 0.19
C32 575.4 [C32 alcohol + 109Ag]+ 1.93 ± 0.36 2.77 ± 0.34 0.52 ±0.26
Fatty acid C18:3 385.3 [C18:3 fatty acid + 107Ag]+ 30.04 ± 2.24 5.40 ± 0.80 -2.48 ± 0.20
C24 475.3 [C24 fatty acid + 107Ag]+ 8.45 ± 0.58 2.99 ± 0.24 -1.50 ± 0.13
C26 503.3 [C26 fatty acid + 107Ag]+ 15.63 ± 1.03 4.17 ± 0.48 -1.91 ± 0.16
C30 559.4 [C30 fatty acid + 107Ag]+ 3.87 ± 0.22 18.68 ± 2.41 2.27 ± 0.17
C32 589.4 [C32 fatty acid + 109Ag]+ 0.80 ± 0.10 4.25 ± 0.68 2.41 ± 0.24
Ketone C29 529.4 [C29 ketone + 107Ag]+ 3.38 ± 0.26 2.22 ± 0.05 -0.60 ± 0.10
Unknown 625.5 [Unk-B2 + 107Ag]+ 6.42 ± 0.39 3.64 ± 0.79 -0.81 ± 0.27
669.5 [Unk-A4 + 107Ag]+ 3.33 ± 0.29 2.14 ± 0.43 -0.64 ± 0.26
108
Table 3(C) Sepal
Class [M + Ag]+ ion species R. I.(Ler) (%) R. I.(cer2) (%) Log2(cer2/Ler)
Alkane C27 487.3 [C27 alkane + 107Ag]+ 1.99 ± 0.16 0.74 ± 0.03 -1.42 ± 0.11
C29 515.4 [C29 alkane + 107Ag]+ 19.28 ± 1.65 2.98 ± 0.17 -2.69 ± 0.12
C31 543.4 [C31 alkane + 107Ag]+ 11.25 ± 1.60 6.91 ± 0.14 -0.70 ± 0.17
Alcohol C26 491.3 [C26 alcohol + 109Ag]+ 1.78 ± 0.13 2.34 ± 0.14 0.39 ± 0.12
C28 519.4 [C28 alcohol + 109Ag]+ 7.64 ± 0.72 5.82 ± 0.27 -0.39 ± 0.12
C30 547.4 [C30 alcohol + 109Ag]+ 6.85 ± 0.74 11.93 ± 1.22 0.80 ± 0.18
Fatty acid C18:3 385.3 [C18:3 fatty acid + 107Ag]+ 5.77 ± 2.00 2.45 ± 0.52 -1.23 ± 0.49
C24 475.3 [C24 fatty acid + 107Ag]+ 2.95 ± 0.56 1.31 ± 0.13 -1.17 ± 0.26
C26 503.3 [C26 fatty acid + 107Ag]+ 4.39 ± 0.91 0.76 ± 0.05 -2.53 ± 0.26
C30 559.4 [C30 fatty acid + 107Ag]+ 4.13 ± 0.33 6.54 ± 0.88 0.66 ± 0.19
Ketone C29 529.4 [C29 ketone + 107Ag]+ 5.07 ± 0.76 1.43 ± 0.29 -1.82 ± 0.30
Unknown 625.5 [Unk-B2 + 107Ag]+ 4.60 ± 0.96 1.61 ± 0.29 -1.52 ± 0.33
669.5 [Unk-A4 + 107Ag]+ 2.21 ± 0.71 0.59 ± 0.13 -1.90 ± 0.47
109
References
(1) Nawrath, C. Current Opinion in Plant Biology 2006, 9, 281-287.
(2) Heredia, A. Biochimica Et Biophysica Acta-General Subjects 2003, 1620, 1-7.
(3) Samuels, L.; Kunst, L.; Jetter, R. Annual Review of Plant Biology 2008, 59.
(4) Koornneef, M.; Hanhart, C. J.; Thiel, F. Journal of Heredity 1989, 80, 118-122.
(5) Rashotte, A. M.; Jenks, M. A.; Ross, A. S.; Feldmann, K. A. Planta 2004, 219, 5-13.
(6) Xia, Y. J.; Nicolau, B. J.; Schnable, P. S. Plant Cell 1996, 8, 1291-1304.
(7) Xia, Y. J.; Nikolau, B. J.; Schnable, P. S. Plant Physiology 1997, 115, 925-937.
(8) Hannoufa, A.; McNevin, J.; Lemieux, B. Phytochemistry 1993, 33, 851-855.
(9) McNevin, J. P.; Woodward, W.; Hannoufa, A.; Feldmann, K. A.; Lemieux, B.
Genome 1993, 36, 610-618.
(10) Jenks, M. A.; Tuttle, H. A.; Eigenbrode, S. D.; Feldmann, K. A. Plant Physiology
1995, 108, 369-377.
(11) Jenks, M. A.; Rashotte, A. M.; Tuttle, H. A.; Feldmann, K. A. Plant Physiology 1996,
110, 377-385.
(12) Jenks, M. A.; Tuttle, H. A.; Feldmann, K. A. Phytochemistry 1996, 42, 29-34.
(13) Rashotte, A. M.; Jenks, M. A.; Nguyen, T. D.; Feldmann, K. A. Phytochemistry 1997,
45, 251-255.
(14) Lai, C.; Kunst, L.; Jetter, R. Plant Journal 2007, 50, 189-196.
(15) Goodwin, S. M.; Rashotte, A. M.; Rahman, M.; Feldmann, K. A.; Jenks, M. A.
Phytochemistry 2005, 66, 771-780.
(16) Rashotte, A. M.; Jenks, M. A.; Feldmann, K. A. Phytochemistry 2001, 57, 115-123.
(17) Dutta, T. K.; Harayama, S. Analytical Chemistry 2001, 73, 864-869.
(18) Kuhn, G.; Weidner, S.; Just, U.; Hohner, G. Journal of Chromatography A 1996, 732,
111-117.
(19) Pruns, J. K.; Vietzke, J. P.; Strassner, M.; Rapp, C.; Hintze, U.; Konig, W. A. Rapid
110
Communications in Mass Spectrometry 2002, 16, 208-211.
(20) Yalcin, T.; Wallace, W. E.; Guttman, C. M.; Li, L. Analytical Chemistry 2002, 74,
4750-4756.
(21) Chen, R.; Li, L. Journal of the American Society for Mass Spectrometry 2001, 12,
367-375.
(22) Sluszny, C.; Yeung, E. S.; Nikolau, B. J. Journal of the American Society for Mass
Spectrometry 2005, 16, 107-115.
111
Figure Captions
Figure 1. Arabidopsis flower samples prepared on a LDI target plate for silver-LDI
mass spectral profiling and imaging.
Figure 2. Positive-ion mode silver-LDI mass spectrum taken from an Arabidopsis
thaliana wild-type (Ler) leaf. Mass spectra from 12674 scanning points on the
leaf were averaged. The inset is the expanded m/z range view which includes
the groups of Ag2+ peaks (triplet around m/z 215.8) and Ag3
+ peaks (quartet
around m/z 322.8). The peak at m/z 215.8 which corresponds to the ion [107Ag
+ 109Ag]+ is always the predominant peak through all mass spectra collected
under our experimental conditions.
Figure 3. Ratios of targeted cuticular wax metabolites in Arabidopsis thaliana wild-type
(Ler) and eceriferum mutant (cer2) leaves. Ratios from silver-LDI MS (left)
and GC-MS (right) analyses were scaled as log2 values. Data points in each
colored box correspond to the compound class indicated in a box. The number
of carbons is indicated for each data point. Detailed ion species assignments
and their relative intensities from silver-LDI MS and amounts from GC-MS
are listed in Table 2. Each silver-LDI MS and GC-MS analysis has 3
replicates and error bars correspond to standard errors are shown.
Figure 4. Chemically selective images of Arabidopsis thaliana wild-type (Ler) whole
flowers. The step size for data collection was set to 50 µm for both x and y
directions. Dimension of the image is 5550 µm high × 4960 µm wide. The
value in each image corresponds to the nominal m/z value of the
corresponding ions. All images were processed as normalized intensities and
112
shown as a percentage of the total ion current. All images were smoothed
linearly. Major ions detected at m/z 529 correspond to [C29 ketone + 107Ag]+.
Ions detected at m/z 515, 543 are mainly silver adduct ions of C29 and C31
alkanes. Ions at m/z 547 are from C30 alcohol, and the image for the peak at
m/z 545 corresponds to the overlapped image of C31 alkane and C30 alcohol as
[C31 alkane + 109Ag]+ and [C30 alcohol + 107Ag]+. Images for silver adduct
ions of C18:3 (at m/z 385), C26 fatty acids (at m/z 503), and an unknown
compound (at m/z 625) are also shown.
Figure 5. (A) Chemically selective images of Arabidopsis thaliana wild-type (Ler) and
cer2 mutant flowers. The step size for data collection was set to 50 µm for
both x and y directions. Dimensions of the images for Ler and cer2 are 5550
µm high × 4960 µm wide and 4100 µm high × 4770 µm wide, respectively.
C29 alkane (at m/z 515 as [C29 alkane + 107Ag]+) showed higher abundances
mainly at sepals and carpels in the wild-type than in the cer2 mutant. The
localization of C29 ketone (at m/z 529 as [C29 ketone + 107Ag]+) at the carpel
was easily recognized in the wild-type but not in the cer2 mutant. Higher
abundances of very long chain fatty acids such as C30 and C32 fatty acids (at
m/z 559 as [C30 fatty acid + 107Ag]+ and at m/z 589 as [C32 fatty acid + 109Ag]+)
were found in the cer2 mutant. (B) 3D presentations of the normalized
intensities of chemically selective images of C30 and C32 fatty acids (at m/z
559 and at m/z 589). The yellow arrows in (A) and (B) are for aligning the
viewing angle. Both fatty acids showed high abundances in the tips of petals.
C30 fatty acid showed high abundance in the carpel but C32 fatty acid did not.
113
Figure 6. Positive-ion mode silver-LDI mass spectrum taken from Arabidopsis thaliana
wild-type (Ler) and cer2 mutant flower parts, (A) carpel, (B) petal, and (C)
sepal (right). Each mass spectrum is the average of each flower part. The peak
intensities were normalized to the reference peak (m/z 215.8 which
corresponds to the ion [107Ag + 109Ag]+ ).
Figure 7. Whole flower cuticular wax profiles of Arabidopsis thaliana wild-type (Ler)
and eceriferum mutant (cer2) by GC-MS. Each bar corresponds to the amount
of a specific cuticular wax and the carbon numbers are indicated on x-axis. All
values are scaled as nmol/g fresh weight with ± standard error. Because of the
relatively high amounts of C29 alkane and ketone, they were scaled by a factor
of 0.2 and 0.1, respectively. For values that exceed the y-axis range, the actual
y-values are indicated adjacent to the corresponding bars.
Figure 8. Ratios of targeted cuticular wax metabolites in Arabidopsis thaliana wild-type
(Ler) and eceriferum mutant (cer2) flower parts (carpel, petal, and sepal) by
silver-LDI MS and whole flowers by GC-MS. Ratios were scaled as log2
values. Data points in each colored box correspond to the compound class
indicated in the box. The number of carbons is indicated for each data point.
Detailed ion species assignments and their relative intensities for the three
flower parts were listed in Table 3. Each flower part has 5 replicates and error
bars correspond to standard errors. Whole flower analysis by GC-MS has 3
replicates and the error bars correspond to standard errors.
Figure 9. Proposed data processing scheme of silver-LDI MS data for fast, high-
throughput mutant screening. Sample plots shown were processed with
114
cuticular wax mass profiles from Arabidopsis thaliana wild-type (Ler) and
eceriferum mutant (cer2) carpels.
115
Whole Flower Imaging
Flower Parts Profiling
Figure 1.
116
300 320 340 360 380 400 420 440 460 480 500 520 540 560 580 600 620 640 660 680 700m/z
2000
4000
6000
8000
10000
12000
14000
16000
18000
Inte
nsity
322.8
363.2
365.3
517.4
503.4545.4
505.4
543.4533.4
573.5
571.5519.4475.4
599.5 627.5487.4531.4489.4 547.4
625.5597.5433.4 601.5
435.5 559.4
655.5585.5459.5 613.5 641.5669.5
200 300 400 600 800m/z
20000
40000
60000
80000
Inte
nsity
322.8
215.8
Figure 2.
117
C27
C29
C31
C33
C35
C26
C28
C30
C32
C34
C24
C26
C30
C32
C29
keto
ne
-3-2
-10
12
3
log 2
(cer
2/Le
r)
Alk
ane
Silv
er-L
DI M
S
Fatt
y ac
id
Alc
ohol
C27
C29
C31
C33
C35
C26
C28
C30
C32
C34C2
4
C26
C30
C32
C29
keto
ne
-6-4
-20
24
6
log 2
(cer
2/Le
r)
Alk
ane
Fatt
y ac
id
Alc
ohol
GC-
MS
Figure 3.
118
m/z529 (10%) m/z515 (5%)
m/z545 (3%) m/z547 (2%)
m/z503 (3%) m/z625 (5%)
m/z543 (3%)
m/z385 (5%)
Ler
Figure 4.
119
m/z
529
(10%
)m
/z51
5 (5
%)
Ler
cer2
m/z
559
(5%
)
Y (
mic
ron)
Y (
mic
ron)
4000
4000
3000
3000
2000
2000
1000
1000
00
bundance
bundance
5.0
05.0
0
0.0
00.0
0
X (
mic
ron)
X (
mic
ron)
4000
4000
3000
3000
2000
2000
1000
1000
00
Y (
mic
ron)
Y (
mic
ron)
4000
4000
3000
3000
2000
2000
1000
1000
00
bundance
bundance
1.0
01.0
0 0.5
00.5
0 0.0
00.0
0
X (
mic
ron)
X (
mic
ron)
4000
4000
3000
3000
2000
2000
1000
1000
00
m/z
559
(5%
)m
/z58
9 (1
%)
(A) (B
)
m/z
589
(1%
)
Figure 5.
120
200 250 300 350 400 450 500 550 600 650 700m/z
0
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
35
40
45
50
Rel
ativ
eA
bund
ance
0
5
10
15
20
25
30
35
40
45
50
531.4
529.4
533.4515.4
363.3 535.3513.4 575.4387.4
415.4 679.5601.4
363.3365.3
387.3
515.4
517.4389.4533.4
627.4475.4 543.5391.4449.3 559.4 669.5585.5
517.5
545.5
363.3
547.5387.4 625.5559.5503.5
391.4 669.5433.4 583.5
Ag2+
Ag3+
Ag2+
Ag3+
Ag2+
Ag3+
Wild-type (Ler)
200 250 300 350 400 450 500 550 600 650 700m/z
0
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
35
40
45
50
0
5
10
15
20
25
30
35
40
45
50
545.4531.4 561.4363.2
517.4387.3 499.3433.3 575.4 679.5650.3
363.2
365.2
559.4
561.5
545.5587.5387.3 515.5
627.5475.4433.4 669.5
545.4
547.4363.1559.4517.4
573.5515.4433.3387.3 627.4 669.5
Ag2+
Ag3+
Ag2+
Ag3+
Ag2+
Ag3+
cer2
(A) Carpel
(C) Sepal
(B) Petal
Figure 6.
121
0
50
100
150
200
C29 Ketone X 1/5 C25 C27 C29 X 1/10 C31
nmol
/g fr
esh
wei
ght
Ketone and Alkanes
Ler cer2
288 327 557 423
0
50
100
150
200
250
C18 C20 C22 C24 C26 C28 C30
nmol
/g fr
esh
wei
ght
Primary Alcohols
Ler cer2
0
50
100
150
200
C16 C18 C18:1 C18:2 C18:3 C22 C24 C26 C28 C30
nmol
/g fr
esh
wei
ght
Free fatty acids
Ler cer2 Figure 7.
122
C29
C31
C28
C30
C32
C18:3
C26
C30
C29 Ketone
-3 -2 -1 0 1 2 3
log2(cer2/Ler)
Alkane
Carpel
Fatty acid
Alcohol
C27
C29
C31
C26
C28
C30
C32
C18:3
C24
C26
C30
C32
C29 Ketone
Unk-B2
Unk-A4
-3 -2 -1 0 1 2 3
log2(cer2/Ler)
Alkane
Petal
Fatty acid
Alcohol
C16C18
C18:1
C18:2C18:3
C22
C24
C26
C28 C30
C25C27
C29C31
C18 C20C22
C24C26
C28
C30
C29 Ketone
-8 -6 -4 -2 0 2 4 6 8
log2(cer2/Ler)
Alkane
GC-MSWhole flower
Fatty acid
Alcohol
C27
C29
C31
C26
C28
C30
C18:3
C24
C26
C30
C29 Ketone
Unk-B2
Unk-A4
-3 -2 -1 0 1 2 3
log2(cer2/Ler)
Alkane
Sepal
Fatty acid
Alcohol
Figure 8.
123
MS
Raw
Dat
a Ex
trac
tion
Conv
ertin
g to
Rel
ativ
e In
tens
ity (R
I) Pr
ofile
s (t
o m/z
215.
8)
Stat
istic
al A
naly
sis
for
Repl
icas
Sim
plify
ing
MS
Dat
a by
El
imin
atin
g Pe
aks
from
13C
or 2 H
(Tak
ing
odd
nom
inal
m
asse
s on
ly)
Plot
ting
Log(
RI(m
utan
t)/R
I(wild
-ty
pe))
Peak
Filt
erin
g by
R.I.
s
(pic
king
pea
ks w
hich
R.I.
ei
ther
from
wild
-typ
e or
fr
om m
utan
t is
gre
ater
th
an 1
.0%
)
Gro
upin
g Fi
ltere
d Pe
aks
Base
d on
Pos
sibl
e O
verl
aps D
ue to
Silv
er
Isot
opes
Sub-
grou
ping
in o
ne g
roup
ba
sed
on D
iffer
ence
s in
RIs
and
Stan
dard
Dev
iatio
ns
Sort
ing
Sub-
grou
ps
Acc
ordi
ng to
The
ir
Abs
olut
e Lo
g Va
lues
List
ing
Poss
ible
Co
mpo
unds
Bas
ed o
n G
C-M
S D
ata
Log 2
(RI(
Mut
ant)
/RI(
Wild
)) P
lot
Filt
erin
gG
roup
ing
RI d
iffer
ence
-3-2-10123
350
400
450
500
550
600
650
Log2 (cer2/Ler)
m/z
cer2
/Ler
Carp
els
-3-2-10123
350
400
450
500
550
600
650
Log2 (cer2/Ler)
m/z
cer2
/Ler
Carp
els
-3-2-10123
350
400
450
500
550
600
650
Log2 (cer2/Ler)
m/z
cer2
/Ler
Carp
els
513 51
551751
9
529 53
153353
5
-3
-2.5-2
-1.5-1
-0.50
510
520
530
540
550
Log2 (cer2/Ler)
m/z
cer2
/Ler
Carp
els
51351
5 517 51
952
752953
1 533 53
5
54354
5547
-10-5051015202530
510
520
530
540
550
RI(wild)-RI(mutant) (%)
m/z
RI(
Ler)
-R
I(ce
r2) f
or c
arpe
l
Sub-
grou
ping
C 29
alka
ne
C 29
keto
ne
Figure 9.
124
CHAPTER 5. COLLOIDAL SILVER LASER
DESORPTION/IONIZATION MASS SPETROMETRY OF STEROLS
AND DIRECT PROBING OF CHOLESTEROL ON ASTROCYTE CELL
MONOLAYER
A paper prepared for submission to Rapid Communications in Mass Spectrometry
Sangwon Cha, Ksenija Jeftinija, Srdija Jeftinija, and Edward S. Yeung
Abstract
By using colloidal silver as a matrix, laser desorption/ionization mass spectrometric
analysis of sterols was performed. Cholesterol and phytosterols were sensitively detected as
silver adduct ions with little or no fragmentation. In-source dehydration difference depending
on hydroxyl group orientation was observed. By spraying colloidal silver homogeneously
onto the cell monolayer, cholesterol levels were probed directly from the cell monolayer
surface. Protocol for assaying relative cholesterol abundances was developed. Based on the
protocol, mass spectrometric assay of free cholesterol abundances depending on methyl beta-
cyclodextrin(MβCD) treatment time was performed and their results were compared with
results by traditional fluorometric cholesterol assay method. LDI MS protocol had simpler
sample preparation and assay steps than the traditional method. Both methods showed the
same trend of decreases in cholesterol level according to MβCD treatment time, but observed
ratios of decreases were higher in silver-LDI protocol than in enzymatic fluorometry.
125
Introduction
Cholesterol is one of major components in cell membrane and is highly localized at
lipid rafts in cell membrane. Cholesterol- and sphingolipids- rich lipid raft microdomains on
cell membrane have numerous functions including organization of neurotransmitter
signaling.1 To explore their function and associated proteins, monitoring changes of
neurotransmitters after disruption of lipid rafts is the common approach.2 Disruption of lipid
raft domain was mainly performed by inhibition of cholesterol synthesis or chemical removal
of cholesterol.2 Therefore, cholesterol level on cell membrane is a very important factor to
understand signal transduction mechanism. For chemical removal of cholesterol,
cyclodextrins which are cyclic oligosaccahrides having a lipophilic central cavity were used
for capturing cholesterol molecules.3, 4 Among those, methyl-ß-cyclodextrin (MβCD) has
known to be the most effective capturing agent for cholesterol.5 In addition, unusual
cholesterol level in diseases such as Alzheimer’s disease6 and atherosclerosis7 were found.
Therefore, probing cholesterol level on cell membrane is also important to understand
diseases.
Free cholesterol and cholesteryl esters from cultured cells are traditionally quantified
by enzymatic fluorometry.8-11 However, this fluorometric assay lacks an internal standard
which could cause miscalculation of cholesterol content.12 In addition, recovery of
cholesterol after extracting and redissolving cellular lipids in various solvents was
incomplete and the ratio of recovery also varied according to solvent compositions used.13
Chromatographic separation techniques, such as thin layer chromatography (TLC)14, gas
chromatography(GC) and high performance liquid chromatography (HPLC)15-19, have been
also utilized for measuring cholesterol amounts from cell cultures or human serum.
By combining chromatographic separation techniques with a variety of mass
spectrometric methodologies, identification as well as quantification of sterols has become
more straightforward. In addition, tandem MS have made it possible to elucidate structures of
126
sterols. GC coupled with electron impact ionization MS has been used for typical analysis of
sterols. GC-MS is sensitive and selective but chemical modification due to high temperature
analysis condition and laborious derivatization steps are disadvantageous. HPLC were
coupled with several ionization techniques. Among those, HPLC with atmospheric pressure
chemical ionization (APCI) MS is the most popular method20-22 whereas standard
electrospray ionization (ESI) method is not efficient ionization methods for sterols23.
Therefore, derivatization of sterols before spraying24 and silver ion coordinated spraying25
were demonstrated for effective ion generation through ESI in analysis of sterols and their
esters.
Differently from ionization techniques mentioned above, beam-induced ionization
techniques such as matrix-assisted laser desorption/ionization MS (MALDI MS) and
seconday ion mass spectrometry (SIMS) have been used off-line for extract analysis.
However, sensitivity of sterols in conventional MALDI MS was poor and most of sterols
were detected as dehydrated molecules which could cause losses of structural information.26
Recently, sterols were derivatized with Girad P hydrazine to Girad P hydrazones, and they
were readily detected by conventional MALDI using α-cyano-4-hydroxycinammic acid as a
matrix.27 Time-of-flight (TOF)- SIMS was also used for identification of cholesteryl ester
from HPLC lipid fractions.12 Because sample was deposited on pre-etched silver target,
cholesteryl esters were detected as silver adduct ions.12
However, beam-based techniques also have a capability of direct sampling from
intact cells or tissues. Cholesterol in animal tissues or cells was directly detected by TOF-
SIMS or MALDI MS28-31. In TOF-SIMS, surface metallization (metal enhanced-SIMS) by
silver or gold enhanced secondary ion yields and metal adduct ions of cholesterol was
predominantly detected in case of silver coating on sample surface.28, 32 By scanning through
cell or animal tissue surfaces, localization information of cholesterol was also obtained by
127
TOF-SIMS.28-30 Recently, relative quantification of cholesterol for individual cells was
successfully demonstrated by combining TOF-SIMS with in situ fluorescence microscopy.33
Recently, we first introduced aqueous silver colloidal solution as a matrix for LDI MS
of various wax compounds.34 Here we extend the use of colloidal silver matrix for analysis of
sterols inspired by silver metallization in TOF SIMS and Ag+ coordination ionspray MS. By
using a colloidal silver matrix, mass spectral profiles for sterols were obtained and in-source
dehydration patterns of them were investigated. In addition, direct probing free cholesterol
level on Astrocyte cell monolayer was performed. Free cholesterol level on cell surfaces was
manipulated by treating cell cultures with methyl-ß-cyclodextrin (MβCD). Relative
cholesterol peak intensities to the reference peak ([107Ag + 109Ag]+) were used for estimating
relative abundance of cholesterol on Astrocyte cell monolayer and results from LDI MS were
compared with those from traditional enzymatic fluorometry.
Experimental Section
Materials and chemicals. Sterol standards were purchased from Steraloids Inc.
(Newport, RI) and plant sterol mixture was obtained from Mattrya LLC (Pleasant Gap, PA).
Colloidal suspension of silver (20 ppm) was purchased from Purest Colloids (Westhampton,
NJ). The Amplex Red Cholesterol Assay Kit was from Molecular Probes (Eugene, OR).
Earle’s Balanced Salt Solution (EBSS), Dulbecco’s Modified Eagle Medium (DMEM) high
glucose, fetal bovine serum (FBS), and trypsin (0.25%) were purchased from Invitrogen
(Carlsbad, CA). Papain was obtained from Sigma-Aldrich (St. Louis, MO). All other
chemicals were obtained from Fisher Scientific (Fairlawn, NJ).
Mass spectrometer. A Thermo Finnigan LTQ linear ion trap mass spectrometer
equipped with vMALDI source (Mountain View, CA) was used for mass spectral analysis.
nitrogen laser pulses (337 nm, maximum energy of 280 µJ/pulse, and maximum frequency of
20 Hz) were introduced through the fiber optic cable with 200µm in diameter. The laser spot
128
size on the sample plate surface was about 100 µm in diameter. The pressure of the sampling
chamber was kept at 0.17 Torr by nitrogen gas flow. All mass spectra were collected in the
positive-ion mode.
Cell culture. Mix neuron-glia cultures from neonatal rat cerebral cortex were
established as described. Cortexes from 3 newborn pups (1-3 days old) were freshly dissected
and tissues were enzymatically treated in 2 mL of papain solution (1.54 mg/mL in EBSS) for
40 min at 37 °C. After washing tissues with EBSS, tissues were treated with trypsin inhibitor
and mechanically dissociated in the DMEM medium (DMEM high glucose : heat-inactivated
FBS : penicillin/streptomycin = 89 : 10 : 1 by volume) using pipettes. Cells were plated in
the DMEM medium in culture flasks and maintained at 37 °C in a humidified 5% CO2/95%
air atmosphere. The cells were maintained by changing the medium every 2-3 days. The
cultures were then shaken overnight (12 hours)at 260 rpm at 37 °C. Cultures enriched in type
I astroglia were obtained by trypsinizing the attachedcells for 3 min. Trypsin was inactivated
by adding 3 mL of the DMEM medium. Astrocytes were plated on poly-L-lysine (100 g/mL;
MW 100,000)-coated glass coverslips or stainless steel plates. All experiments were
performed on cells that have been in culture for 1 – 3 days after re-planting.
In vitro fluorometric assay of cholesterol. Astrocyte cells were seeded in a 96-well
microplate at a density of 5 × 104 cells per well. Seeded cells were incubated in the DMEM
medium for 24 hours. The cells were treated with 100 µL of 10mM MβCD solution with
various time periods (0 to 60 minutes). After treatment, cells were washed with KPBS buffer
and cellular lipids were extracted from the cells by adding 100 µL of hexane:2-propanol (3:2,
v/v) solution. First extracts were transferred to the new 96-well microplate. The cellular
lipids were re-extracted with an additional 50 µL of hexane:2-propanol (3:2, v/v) solution
and these second extracts were transferred and merged to the first extracts. Solvents in
extracts were evaporated under vacuum and cellular lipids were resuspended in 150 µL of
phosphate buffer containing 0.1M potassium phosphate, pH 7.4, 50 mM NaCl, 5mM cholic
129
acid, and 0.1% Triton X-100 for 12 hours with shaking. Cellular free cholesterol content was
quantified by using an Amplex Red Cholesterol Assay kit (Molecular Probes, Eugene, OR)
without cholesterol esterase. Briefly, cholesterol in the cellular lipid extracts is oxidized by
cholesterol oxidase to produce H2O2. In the presence of horseradish peroxidase, 10-acetyl-
3,7-dihydoxyphenoxazine (Amplex Red) reacts with H2O2 and this reaction generates
fluorescent resorufin quantitatively. If cholesterol esterase is present, cholesterol esters can
be hydrolyzed to cholesterol. Fluorescence was measured with the fluorescence microplate
reader (Spectra Max Gemini XS, Molecular Devices, Sunnyvale, CA) using excitation at 538
nm and emission at 590 nm. Cell residues were used for cellular protein quantification by the
micro BCA protein assay kit (Pierce, Rockford, IL). The amount of cholesterol was
normalized by the amount of cellular protein.
LDI mass spectrometric assay of cholesterol. Astrocyte cells were planted on a
poly-L-lysine-coated stainless steel plate. Four spots of cell monolayer were formed on a
stainless steel plate (25 mm × 75 mm). Each spot of cell monolayer had about 5 × 104 cells in
an area of about 1.5 cm2. After planting, cells were incubated in the DMEM medium for 24
hours. The cells then were treated with MβCD solution with various time periods (0 to 60
minutes). Immediately after MβCD treatment, all cell monolayers were fixed with 4% of
paraformaldehyde solution for 30 min and transferred to KPBS buffer solution. Before
spraying colloidal silver, cell monolayers were dried under moderate vacuum (~50 Torr).
Spraying of colloidal silver solution were performed by the home-built spraying
device which is the combination of a computer-controlled syringe pump with a 500 μL
syringe (Kloehm LTD., Las Vegas, NV), MicroFlow PFA-ST nebulizer (Elemental Scientific
Inc., Omaha, NE) with a 0.25 mm i.d. sample uptake capillary, and a helium gas cylinder.
Spraying action was controlled by the supplied software “WinPump®” by Kloehm. Six sets
of spraying-drying processes were performed for each spot of cell monolayer with 100
μL/min flow rate and 12.5 μL of colloidal silver solution per spraying. Because the area
130
covered by colloidal silver on the sample plate was about 4.5 cm2 which was larger than the
spot of cell monolayer (~1.5 cm2), spraying action was performed without moving the
sample plate.
Based on optical images of cell monolayer spots which were acquired in the mass
spectrometer, rastering areas were selected and scanned with a 100 μm movement. Five laser
shots were used to obtain the mass spectrum for one scanning point. For one spot of cell
monolayer, 485 mass spectra from different scanning points were collected and then
averaged. From averaged mass spectra, intensity values of cholesterol silver adduct ion peaks
(at m/z 493.5 and 495.5) were extracted and normalized as relative intensities to the reference
peak at m/z 215.8 ([107Ag + 109Ag]+). These relative intensities were used for calculating
abundances of free cholesterol in cell monolayer. Four replicas of cell monolayers were used
for each MβCD treatment.
Results and Discussion
Silver-LDI mass spectral profiles of sterols. In silver-LDI MS, cholesterol (5-
cholesten-3β-ol) was mostly detected as silver adduct ions at m/z 493-496 and showed a
quartet of peaks with two major peaks because of the two silver isotopes ([M + 107Ag]+ at m/z
493 and [M + 109Ag]+ at m/z 495). Detection limit for cholesterol was ~ 50 pg per sample
spot. However, for high concentration of cholesterol standards (> ~50 ng/sample spot),
dehydrated cholesterol silver adduct ions ([M – H2O + Ag]+) were also detected at m/z 475-
478 with a constant laser fluence (Figure 1(a)). In Figure 1, one interesting observation was
that dehydration ions were not found in the mass spectrum of its isomer, 5-cholesten-3α-ol
(Figure 1(b)). In addition, with high laser fluence, dehydroxylated silver adduct ions ([M –
OH + Ag]+) were observed for 5-cholesten-3β-ol but not for 5-cholesten-3α-ol. This
difference could not be recognized in conventional MALDI MS because only dehydrated
ions were observed for both alpha- and beta- hydroxyl group positions. The explanation for
131
this difference in dehydroxylation or dehydration has not been reported. As shown in Figure
3, an efficient orbital overlap between beta-hydroxy and adjacent alkene group may have
stabilizing effects of the intermediates in either dehydroxyation or dehydration (Figure 2). In
addition, the other possible explanation is that beta-hydroxy and adjacent alkene group are
geometrically aligned well with the hydrogen orbital group at the allylic position and
therefore this may facilitate syn elimination of water (Figure 2).
Figure 3 shows silver-LDI mass spectrum of major phytosterol mixture. As shown in
Figure 3, sterols which are 2 Da different in molecular weight were overlapped due to two
stable silver isotopes. In other words, the peak at m/z 507 was from both [Brassicasterol + 109Ag]+ and [Campesterol + 107Ag]+. Similarly, [Stigmasterol + 109Ag]+ and [β-Sitosterol + 107Ag]+ are associated to the peak at m/z 521. However, their intensities could be individually
measured by choosing the other silver isotope adduct ion (Figure 3).
LDI mass spectrometric assay of cholesterol on cell monolayer. With colloidal
silver spraying on cell monolayer, surface cholesterol was directly probed. Silver-LDI mass
spectrum directly taken from an Astrocyte cell monolayer spot was shown in Figure 4. As
shown in Figure 4, silver adduct ions of cholesterol and dehydrated cholesterol were readily
detected at m/z 493 - 496 and at m/z 475 – 478 directly from an Astrocyte cell monolayer spot.
For obtaining mass spectral profiles directly from the cell monolayer on poly-L-lysine coated
stainless steel surface, about 20 % ~ 30 % higher laser intensity was needed compared to that
needed for standard sterol or sterol extract analysis. This may cause higher ratio of
dehydrated cholesterol ions. The peak corresponding to the silver dimer ion at m/z 215.8
([107Ag + 109Ag]+) was predominant under our silver spraying condition and this was
consistent through mass spectra from all individual scanning points on both control and
MβCD-treated cell monolayers. In addition, the relative intensities of cholesterol on one cell
monolayer with respect to the peak intensity at m/z 215.8 were consistent with a less than
10 % of relative standard deviation. Therefore, the fractional peak intensities for peaks at m/z
132
475, 477, 493, and 495 to the peak intensity at m/z 215.8 were extracted from the averaged
mass spectrum and summed for comparing relative abundances of cholesterol among cell
monolayer spots.
It should be noted that mass spectra from all scanning points showed a variation in
terms of total ion current if cell monolayer was prepared on the poly-L-lysine coated cover
slip glass. In addition, intensities from cholesterol were also about one order of magnitude
lower than those from the cell monolayer on the stainless steel plate even if 50% higher laser
intensity was used. Therefore, all cell monolayer spots were prepared on poly-L-lysine
coated stainless steel plates for obtaining relative abundance information.
Figure 5 shows relative cholesterol levels of control and MβCD-treated cell
monolayer spots which were assayed by both enzymatic fluorometry and silver-LDI MS.
From Figure 5, several features from silver-LDI MS could be summarized and compared
with enzymatic fluorometry results. First, the silver-LDI MS method successfully probed
decreased cholesterol levels on MβCD treated cell monolayer spots by estimating cholesterol
abundances with relative peak intensities. In addition, relative peak intensities of cholesterol
from four equally-treated cell monolayer spots showed less than 8 % of intensity variations.
Second, degrees of decreased cholesterol levels by MβCD treatment were found to be
consistent in independently cultured cell batch A and B by both assay methods. Third, the
silver-LDI MS method showed the larger decreases of cholesterol levels than the enzymatic
fluorometry method in the same treatment. It may be because sampling region by silver-LDI
MS is mostly limited to cell surfaces where were affected by MβCD treatment whereas
enzymatic fluorometry method probe cholesterol level from total cellular lipid extracts. This
observation may suggest that the effect of MβCD treatment on cell monolayer could be
underestimated by enzymatic fluorometry method.
133
Conclusions
LDI MS of sterols using colloidal silver as a matrix were performed and cholesterol
levels on cell monolayer spots were demonstrated. With low detection limit, cholesterol and
phytosterol standards were mainly detected as intact silver adduct ions, and a difference in
in-source dehydration between cholesterol and its isomer was observed which could not be
recognized when using conventional MALDI matrixes. By performing homogeneous and
reproducible spraying of colloidal silver solution with the micronebulizer-based spraying
device, consistent mass spectral profiles could be obtained through surfaces of Astrocyte cell
monolayer spots. Sample preparation and assay procedure of the silver-LDI MS method was
much simpler and faster than the traditional enzymatic fluorometry method because it didn’t
need any lipid extraction and enzymatic reaction step. Sampling depth by LDI-MS is thought
to be limited to the outer cell membrane surface but is still ambiguous. Further investigations
on sampling depth according to laser fluence, cell population, and planting surface
characteristics are needed.
Ackowledgement
E.S.Y. thanks the Robert Allen Wright Endowment for Excellence for support. The
Ames Laboratory is operated for the US Department of Energy by Iowa State University
under Contract No. W-7405-Eng-82. This work was supported by the Director of Science,
Office of Basic Energy Sciences, Divisions of Chemical Sciences and Biosciences.
134
References
(1) Tsui-Pierchala, B. A.; Encinas, M.; Milbrandt, J.; Johnson, E. M. Trends in
Neurosciences 2002, 25, 412-417.
(2) Allen, J. A.; Halverson-Tamboli, R. A.; Rasenick, M. M. Nature Reviews
Neuroscience 2007, 8, 128-140.
(3) Wolozin, B. Proceedings of the National Academy of Sciences of the United States of
America 2001, 98, 5371-5373.
(4) Uekama, K.; Hirayama, F.; Irie, T. Chemical Reviews 1998, 98, 2045-2076.
(5) Abulrob, A.; Tauskela, J. S.; Mealing, G.; Brunette, E.; Faid, K.; Stanimirovic, D. J
Neurochem 2005, 92, 1477-1486.
(6) Simons, K.; Ehehalt, R. Journal of Clinical Investigation 2002, 110, 597-603.
(7) Lusis, A. J. Nature 2000, 407, 233-241.
(8) Allain, C. C.; Poon, L. S.; Chan, C. S. G.; Richmond, W.; Fu, P. C. Clinical
Chemistry 1974, 20, 470-475.
(9) Amundson, D. M.; Zhou, M. J. Journal of Biochemical and Biophysical Methods
1999, 38, 43-52.
(10) Gamble, W.; Vaughan, M.; Kruth, H. S.; Avigan, J. Journal of Lipid Research 1978,
19, 1068-1070.
(11) Heider, J. G.; Boyett, R. L. J. Lipid Res. 1978, 19, 514-518.
(12) Cullen, P.; Fobker, M.; Tegelkamp, K.; Meyer, K.; Kannenberg, F.; Cignarella, A.;
Benninghoven, A.; Assmann, G. Journal of Lipid Research 1997, 38, 401-409.
(13) Cullen, P.; Tegelkamp, K.; Fobker, M.; Kannenberg, F.; Assmann, G. Anal Biochem
1997, 251, 39-44.
(14) Schmitz, G.; Assmann, G.; Bowyer, D. E. Journal of Chromatography 1984, 307, 65-
79.
135
(15) Araki, N.; Horiuchi, S.; Rahim, A.; Takata, K.; Morino, Y. Analytical Biochemistry
1990, 185, 339-345.
(16) Contreras, J. A.; Castro, M.; Bocos, C.; Herrera, E.; Lasuncion, M. A. Journal of
Lipid Research 1992, 33, 931-936.
(17) Kuo, J. C.; Yeung, E. S. Journal of Chromatography 1982, 229, 293-300.
(18) Carroll, R. M.; Rudel, L. L. J. Lipid Res. 1981, 22, 359-363.
(19) Greenspan, M. D.; Lo, C. Y.; Hanf, D. P.; Yudkovitz, J. B. J. Lipid Res. 1988, 29,
971-976.
(20) Raith, K.; Brenner, C.; Farwanah, H.; Muller, G.; Eder, K.; Neubert, R. H. H. Journal
of Chromatography A 2005, 1067, 207-211.
(21) Tian, Q.; Failla, M. L.; Bohn, T.; Schwartz, S. J. Rapid Commun Mass Spectrom 2006,
20, 3056-3060.
(22) Palmgren, J. J.; Toyras, A.; Mauriala, T.; Monkkonen, J.; Auriola, S. Journal of
Chromatography B-Analytical Technologies in the Biomedical and Life Sciences
2005, 821, 144-152.
(23) Chen, B. H.; Chen, Y. C. Journal of Chromatography A 1994, 661, 127-136.
(24) Griffiths, W. J.; Wang, Y.; Alvelius, G.; Liu, S.; Bodin, K.; Sjovall, J. J Am Soc Mass
Spectrom 2006, 17, 341-362.
(25) Seal, J. R.; Havrilla, C. M.; Porter, N. A.; Hachey, D. L. Journal of the American
Society for Mass Spectrometry 2003, 14, 872-880.
(26) Schiller, J.; Zschornig, O.; Petkovic, M.; Muller, M.; Arnhold, J.; Arnold, K. Journal
of Lipid Research 2001, 42, 1501-1508.
(27) Wang, Y. Q.; Hornshaw, M.; Alvelius, G.; Bodin, K.; Liu, S. Y.; Sjovall, J.; Griffiths,
W. J. Analytical Chemistry 2006, 78, 164-173.
(28) Nygren, H.; Malmberg, P.; Kriegeskotte, C.; Arlinghaus, H. F. Febs Letters 2004, 566,
291-293.
136
(29) Nygren, H.; Johansson, B. R.; Malmberg, P. Microscopy Research and Technique
2004, 65, 282-286.
(30) Altelaar, A. F. M.; Klinkert, I.; Jalink, K.; de Lange, R. P. J.; Adan, R. A. H.; Heeren,
R. M. A.; Piersma, S. R. Analytical Chemistry 2006, 78, 734-742.
(31) Jackson, S. N.; Wang, H. Y. J.; Woods, A. S. Analytical Chemistry 2005, 77, 4523-
4527.
(32) Inoue, M.; Murase, A. Surface and Interface Analysis 2005, 37, 1111-1114.
(33) Ostrowski, S. G.; Kurczy, M. E.; Roddy, T. P.; Winograd, N.; Ewing, A. G. Anal.
Chem. 2007, 79, 3554-3560.
(34) Sluszny, C.; Yeung, E. S.; Nikolau, B. J. Journal of the American Society for Mass
Spectrometry 2005, 16, 107-115.
(35) Hishiya, T.; Asanuma, H.; Komiyama, M. J. Am. Chem. Soc. 2002, 124, 570-575.
137
Figure Captions
Figure 1. Colloidal silver-LDI mass spectra of cholesterol (5-cholesten-3β-ol) and 5-
cholesten-3α-ol. Sample loading was 100 ng/spot for both compounds. With
the same laser intensity, dehydrated peak [M – H2O +Ag]+ was observed for
5-cholesten-3β-ol but not for 5-cholesten-3α-ol.
Figure 2. Possible explanation for the difference in in-source dehydration of 5-
cholesten-3β-ol and 5-cholesten-3α-ol.
Figure 3. Colloidal silver-LDI mass spectrum of plant sterol mixture solution. Sample
loading was 10 ng of total plant sterols/spot. Four sterol compounds were not
in equal amount in the mixture. The inset is the enlarged view of the region
m/z 500-530 where plant sterols were detected with their structures. Peaks
with the asterisk are from unknown sample plate contaminants.
Figure 4. Colloidal silver-LDI mass spectral profile directly taken from the Astrocyte
cell monolayer spot. Silver adduct ions for cholesterol and dehydrated
cholesterol were detected at m/z 493 - 496 and at m/z 475 – 478. Peaks in the
m/z 220 – 600 were magnified with factor of two. Peaks with the asterisk are
from unknown sample plate contaminants.
Figure 5. Effect of MβCD treatment on Astrocyte cells which was assayed by both
silver- LDI MS and enzymatic fluorometry method. Cholesterol levels in
silver-LDI MS were based on sum of relative intensities of peaks from
cholesterol and its fragment. Cholesterol levels in enzymatic fluorometry were
138
based on the cholesterol content which was normalized by protein content in
cells. Silver- LDI MS had 4 replicas for each treatment and enzymatic
fluorometry had 8 replicas for each treatment. Cholesterol levels were
indicated as relative abundances to the cholesterol level of control (without
MβCD treatment). Cholesterol assay were performed on two independently
cultured batches A and B. Error bars correspond to ± relative standard
deviation values.
139
450 460 470 480 490 500 510 520m/z
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Rel
ativ
eAb
unda
nce
493.36
495.36
475.34 477.34
5-cholesten-3β-ol(cholesterol)
5-cholesten-3α-ol
[M + Ag]+
[M - H2O + Ag]+
(a)
(b)
Figure 1.
140
HOH H
HH
HOH H
HH
H
HO
LUMO
syn-elimination
delocalizedcarbocation
δ−
CH3H3C
5-Cholesten-3β-ol 5-Cholesten-3α-ol
Figure 2.
141
200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520m/z
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Rel
ativ
eA
bund
ance
500 505 510 515 520 525 530m/z
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
[Brassicasterol + 107Ag]+
[Campesterol + 109Ag]+
[Stigmasterol + 107Ag]+
[ß-Sitosterol + 109Ag]+
505.33
523.33
521.33
519.42
509.25
507.33
HOH H
HH
HOH H
HH
HOH H
HH
HOH H
HH
Brassicasterol
Campesterol
Stigmasterol
ß-Sitosterol
Rel
ativ
eA
bund
ance
Ag2+
Ag3+
*
Figure 3.
142
200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540 560 580 600m/z
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Rel
ativ
eA
bund
ance
x2
*
*
*
*
Ag2+
Ag3+
[M + Ag]+
[M - H2O + Ag]+
234.08
309.33
421.33
391.33363.33291.33553.33
Figure 4.
143
0.0
50.0
100.0
0 min (control) 10 min 30 min
Fluorometry 100.0 78.0 67.9
Silver-LDI MS 100.0 64.6 50.0
Rela
tive
Abu
ndan
ce
Culture - A
0.0
50.0
100.0
0 min (control) 10 min 30 min
Fluorometry 100.0 76.1 70.5
Silver-LDI MS 100.0 67.0 52.5
Rela
tive
Abu
ndan
ce
Culture - B
Figure 5.
144
CHAPTER 6. GENERAL CONCLUSIONS
In this dissertation, laser desorption/ionization (LDI) mass spectrometry (MS) and
imaging mass spectrometry of small molecules was investigated with various interesting
biological systems. With use of alternative matrixes such as colloidal graphite and colloidal
silver solution, small molecules which were not detected very well with conventional
MALDI method were successfully detected.
With colloidal graphite as a matrix, cerebrosides were readily detected and imaged
from rat brain tissues whereas this class of compounds was usually suppressed in
conventional MALDI MS spectra of lipid mixture by the signals from phosphatidylcholine
species. For isobaric ions present in lipid mass spectral profiles, imaging tandem mass
spectrometry was performed and overlapped localization information was successfully
separated. In addition, GALDI MS was applied to plant materials such as fruit samples and
Arabidopsis thaliana plants which is one of the most important plant models. Various classes
of secondary metabolites such as small organic acids, oligosaccharides, long chain fatty acids,
and flavonoids were detected with high sensitivity. Especially, by imaging of Arabidopsis
flower petals, the heterogeneous distribution of flavonoid species in a single flower organ
was discovered for the first time by GALDI imaging MS. This discovery has a possibility of
expanding understandings of functional genomics studies of the flavonoid biosynthetic
pathway.
With colloidal silver matrix, cuticular wax compounds were directly profiled and
imaged from Arabidopsis thaliana flower surfaces. This wax metabolite profiling of flower
parts has never been studied by the traditional GC-MS protocol. Reproducible mass profiles
from both wild-type and mutants were obtained by using the homogeneous colloidal silver
spraying technique. Comparison of relative abundances of wax compounds through wild-
type and mutants by silver LDI MS provided the idea of mutant gene product functions in
145
wax biosynthesis pathway and showed good agreement with traditional GC-MS results. In
addition, cholesterol levels were directly probed from Asctrocyte cell monolayers by using
silver LDI MS.
Overall, alternative matrix-based direct profiling techniques have much simpler
sample preparation steps and shorter analysis time than extraction- and separation – based
traditional methods. Much smaller sample amount required for the analysis is also the big
advantage of direct profiling LDI MS. In addition, compared to organic acid-based MALDI
matrixes, alternative matrixes produce generally cleaner background in the low mass region
and more homogeneous sample surfaces can be generated which is suitable for imaging MS
applications. However, identification of small molecules from complete biosystems by LDI
MS is still hard to achieve for some cases and there is a high possibility of isobaric ion
overlaps in low mass region profiles. In addition, the sampling depth by LDI MS is thought
to be limited to very outer surface but still ambiguous. Therefore, further systematic studies
on separating isobaric ions based on tandem MS or high mass resolution MS or ion mobility
MS are needed, and investigation of the sampling depth according to matrix application, laser
properties, and sample materials is must.
146
APPENDIX 1. SUPPORTING FIGURES FOR CHAPTER 2
This supporting information is published online at http://pubs.acs.org*
Sangwon Cha and Edward S. Yeung
Figure Captions
Figure S-1. High-vacuum (HV)-MALDI and HV-GALDI mass spectra of standard lipid
mixtures of Batch 1. Solutions of Batch 1 are composed of a constant
concentration of total glucosylceramides (GlcCers) (0.5 mg/ml) and various
concentrations of phosphatidylcholine 32:0 (PC 32:0) (from 0.005 mg/ml to
0.25 mg/ml). Ratios on the right side of the figure represent weight ratios of
PC 32:0 to total GlcCers in each mixture. Relative abundances for the major
mass peaks are listed in Table 2.
Figure S-2. HV-MALDI and HV-GALDI mass spectra of standard lipid mixtures of Batch
2. Solutions of Batch 2 are composed of a constant concentration of PC 32:0
(0.5 mg/ml) and various concentrations of total GlcCers (from 0.005 mg/ml to
0.25 mg/ml). Ratios on the right side of the figure represent weight ratios of
___________________________________________________________________________
* http://pubs3.acs.org/acs/journals/supporting_information.page?in_manuscript=ac062251h
Copyright © 2007 American Chemical Society
147
PC 32:0 and total GlcCers in each mixture. Relative abundances for the major
mass peaks are listed in Table 2.
Figure S-3. Intermediate pressure (IP)-MALDI and IP-GALDI mass spectra of standard
lipid mixtures of Batch 1. Solutions of Batch 1 are composed of a constant
concentration of total GlcCers (0.5 mg/ml) and various concentrations of PC
32:0 (from 0.005 mg/ml to 0.25 mg/ml). Ratios on the right side of the figure
represent weight ratios of PC 32:0 to total GlcCers in each mixture. Relative
abundances for the major mass peaks are listed in Table 2.
Figure S-4. IP-MALDI and IP-GALDI mass spectra of standard lipid mixtures of Batch 2.
Solutions of Batch 2 are composed of the constant concentration of PC 32:0
(0.5 mg/ml) and various concentrations of total GlcCers (from 0.005 mg/ml to
0.25 mg/ml). Ratios on the right side of the figure represent weight ratios of
PC 32:0 and total GlcCers in each mixture. Relative abundances for major
mass peaks are listed in Table 2.
Figure S-5. Mass spectrum of rat brain tissue in the negative-ion mode. This mass
spectrum is the average of mass spectra from all rastering points (10,817
points) on the rat brain tissue.
Figure S-6. First generation product ion spectra of ions at m/z 844.50. NL 59, NL 124, NL
183, and NL 256 could correspond to the loss of trimethylamine, ethyl
phosphate, phosphocholine head group, and palmitic acid from [PC 38:6 +
K]+, respectively. NL 43, NL 284, and NL 304 could correspond to the loss of
148
aziridine, stearic acid (18:0), and arachidonic acid (20:4) from [PE 38:4 + 2K
– H]+, respectively.
149
HV HIDLAM- V IDLAG- reCclG:CP s
2:1
4:1
02:1
001:1
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
26.437
96.608
65.65726.537
17.028
95.757 56.12896.327
009068028087047007)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
06.437
75.657
06.537
56.608
46.02875.75716.227
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
26.437
07.608
85.657
17.02856.227
36.537
27.708
85.75737.438
86.32717.297
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
93.608
83.437
04.028
14.227
04.12824.708
44.43824.08744.667
83.977 14.538
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
95.608
16.028
74.227
06.708 26.438
94.327 65.877 85.297
56.538
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
94.608
15.028
14.227
94.708
25.438
25.12814.327 84.297
35.538
74.877
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
96.608
17.02855.227
96.70847.438
55.327
56.297
56.877 37.53807.808
55.427
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
36.608
56.028
35.227 86.438
46.70856.128
16.29735.327 86.53806.87756.808
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
26.437
96.608
65.65726.537
17.028
95.757 56.12896.327
009068028087047007)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
06.437
75.657
06.537
56.608
46.02875.75716.227
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
26.437
07.608
85.657
17.02856.227
36.537
27.708
85.75737.438
86.32717.297
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
93.608
83.437
04.028
14.227
04.12824.708
44.43824.08744.667
83.977 14.538
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
26.437
96.608
65.65726.537
17.028
95.757 56.12896.327
009068028087047007)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
06.437
75.657
06.537
56.608
46.02875.75716.227
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
26.437
07.608
85.657
17.02856.227
36.537
27.708
85.75737.438
86.32717.297
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
93.608
83.437
04.028
14.227
04.12824.708
44.43824.08744.667
83.977 14.538
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
95.608
16.028
74.227
06.708 26.438
94.327 65.877 85.297
56.538
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
94.608
15.028
14.227
94.708
25.438
25.12814.327 84.297
35.538
74.877
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
96.608
17.02855.227
96.70847.438
55.327
56.297
56.877 37.53807.808
55.427
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
36.608
56.028
35.227 86.438
46.70856.128
16.29735.327 86.53806.87756.808
Figure S- 1.
150
HV HIDLAM- V IDLAG- reCclG:CP s
1:2
1:4
1:02
1:001
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
45.437
45.537
15.657
15.75745.63765.227
16.60825.857 55.02866.327 85.87709.007 85.29785.737 16.438
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
95.437
95.537
55.657
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
95.437
95.537
55.657
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
85.437
85.537
45.657
009068028087047007)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
57.608
36.227
97.028
77.708
18.438
87.12847.29737.877
56.327
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
07.608
57.028
16.227
37.438
47.538
27.297 37.70886.877 17.12846.657
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
85.657
65.757
95.847
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
85.657
36.00795.757
56.60836.43706.107
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
45.437
45.537
15.657
15.75745.63765.227
16.60825.857 55.02866.327 85.87709.007 85.29785.737 16.438
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
95.437
95.537
55.657
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
95.437
95.537
55.657
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
85.437
85.537
45.657
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
45.437
45.537
15.657
15.75745.63765.227
16.60825.857 55.02866.327 85.87709.007 85.29785.737 16.438
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
95.437
95.537
55.657
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
95.437
95.537
55.657
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
85.437
85.537
45.657
009068028087047007)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
57.608
36.227
97.028
77.708
18.438
87.12847.29737.877
56.327
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
07.608
57.028
16.227
37.438
47.538
27.297 37.70886.877 17.12846.657
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tens
ity
85.657
65.757
95.847
009068028087047007
)z/m(ssaM
0
01
02
03
04
05
06
07
08
09
001
%In
tensi
ty
85.657
36.00795.757
56.60836.43706.107
Figure S- 2.
151
680 700 720 740 760 780 800 820 840 860 880 900m/z
0
20
40
60
80
1000
20
40
60
80
1000
20
40
60
80
100
Re
lati
ve
Ab
un
da
nce
0
20
40
60
80
100734.50
756.50
699.08 806.66 820.66722.66 834.75792.66780.58 874.83766.50682.50 750.66712.91 848.75 861.00 896.00734.50
756.41
699.08806.66
820.66722.58 834.66780.50 792.66766.50 874.83682.50 709.25 752.50 847.16 861.00 889.16
699.08 734.41
720.08806.66
820.66722.58 756.41834.75766.50 780.58 874.83682.41 792.66778.66
712.91 752.50685.25847.25 860.91 897.00888.91
734.41699.08
806.66
820.66722.58 756.41
766.50 834.75780.58682.50 792.58778.66 874.83
752.50709.33691.25 847.25889.08857.08
680 700 720 740 760 780 800 820 840 860 880 900m/z
0
20
40
60
80
1000
20
40
60
80
1000
20
40
60
80
100
Re
lati
ve
Ab
un
da
nce
0
20
40
60
80
100806.66
820.66
722.58 756.50834.66
792.66697.41778.66
750.58734.50 848.66694.58 708.58 772.50 863.50 892.50880.41806.66
820.66
722.66834.66
792.66756.50778.66
697.50750.66
848.66734.50694.58 708.66 761.41 863.50 891.41880.41806.66
820.66
722.66
834.66
792.66778.66
750.66848.66733.50697.50 708.66 761.50 863.50689.50 891.41879.41
806.66
820.66
722.66834.66
792.66778.66
750.66 848.66733.41708.66694.58 761.50 863.50 891.50879.50
IP-MALDI IP-GALDI PC: GlcCers
1:2
1:4
1:20
1:100
Figure S- 3.
152
IP-MALDI IP-GALDI PC: GlcCers
2:1
4:1
20:1
100:1
680 700 720 740 760 780 800 820 840 860 880 900m/z
0
20
40
60
80
1000
20
40
60
80
1000
20
40
60
80
100
Re
lati
ve
Ab
un
da
nce
0
20
40
60
80
100734.50
756.50
699.08806.66722.66 820.75 874.83834.75763.25682.50 780.66 792.66709.33 847.33751.25 857.16 889.16
734.50
756.50
699.08 806.58783.41722.66 763.25 874.83835.16820.66709.33682.50 748.50 847.25792.75 856.16 889.25734.50
756.50
699.08 806.50783.41763.25 835.25 874.91709.33685.25 748.50721.25 847.33 856.16823.25 899.25791.16 889.25734.50
756.50
697.41 783.41 806.50763.25 835.25709.33685.25 847.25 874.91751.16721.33 823.25 857.16791.16 889.16
680 700 720 740 760 780 800 820 840 860 880 900m/z
0
20
40
60
80
1000
20
40
60
80
1000
20
40
60
80
100
Re
lati
ve
Ab
un
da
nce
0
20
40
60
80
100756.50
806.66
820.66
697.41834.66
722.66734.50
792.66748.50 778.66
772.50 848.66713.50694.50 863.50 892.41880.33756.50
697.41
806.66820.66
734.50 748.50 834.66722.66792.66778.66713.50693.50 848.66761.41 863.50 892.41880.33
756.50
697.41
734.50 748.50 806.58820.66693.50 835.50705.41 719.50 761.41 779.50 863.50792.66 891.41849.41 877.41
756.50
697.41
734.50748.50
806.50693.50 719.50705.41 772.50 796.41 835.50 863.41819.33 887.91849.41 877.41 895.41
Figure S- 4.
153
800 810 820 830 840 850 860 870 880 890 900 910 920 930 940 950m/z
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Rel
ativ
eA
bund
ance
888.58
890.58
806.50 885.41
906.58878.58822.50862.58
857.33834.25876.50826.16816.08 850.16
902.50 910.08 932.08926.08 938.08
Figure S- 5.
154
450 500 550 600 650 700 750 800 850 900m/z
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Rel
ativ
eA
bund
ance
844.50
785.
21 (N
L 59
)
560.
26 (N
L 28
4)
801.
33 (N
L 43
)
540.
32 (N
L 30
4)
588.
29 (N
L 25
6)
661.
18 (N
L 18
3)
720.
50 (N
L 12
4)
Figure S- 6.
155
APPENDIX 2. DIRECT PROFILING AND MS IMAGING OF SMALL
METABOLITES FROM FRUITES BY COLLOIDAL GRAPHITE-
ASSISTED LASER DESORPTION/IONIZATION MASS
SPECTROMETRY
A paper published in Analytical Chemistry*
Hui Zhang, Sangwon Cha, Edward S. Yeung
Abstract
Due to a high background in the low mass region, conventional MALDI is not as
useful for detecting small molecules (molecular weights < 500 Da) as it is for large ones.
Also, spatial inhomogeneity that is inherent to crystalline matrixes can degrade resolution in
imaging mass spectrometry (IMS). In this study, colloidal graphite was investigated as an
alternative matrix for laser desorption/ionization (GALDI) in IMS. We demonstrate its
advantages over conventional MALDI in the detection of small molecules such as organic
acids, flavonoids and oligosaccharides. GALDI provides good sensitivity for such small
molecules. The detection limit of fatty acids and flavonoids in negative ion mode are in low
femtomoles range. Molecules were detected directly and identified by comparing the MS and
MS/MS spectra with those of standards. Various fruits were chosen to evaluate the practical
utility of GALDI since many types of small molecules are present in them. Distribution of
these small molecules in the fruit was investigated by using IMS and IMS/MS.
___________________________________________________________________________
* Reprint with permission from Analytical Chemistry 2007, 79(17), 6575-6584.
Copyright © 2007 American Chemical Society
156
Introduction
Matrix-assisted laser desorption/ionization mass spectrometry (MALDI MS) has been
extensively used for the analysis of large molecules such as proteins1 and synthetic
polymers.2 The key feature of MALDI MS are its soft ionization characteristic and simplified
spectra as mostly singly charged species are generated. Compared to electrospray ionization
(ESI) MS, MALDI bears other advantages such as better tolerance to interference from salts
and buffers and simpler sample preparation. MALDI has also proven to be very useful for the
analysis of medium-size molecules (500-10 kDa) such as peptides,3 oligonucleotides,4 and
oligo-saccharides.5 However, the analysis of small molecules (< 500 Da) by conventional
MALDI MS is far less successful than that of larger molecules because the analyte ions are
strongly interfered with or are suppressed by the matrix-related ions that are predominant at
the low m/z range.
Different approaches have been employed in MALDI MS to minimize the
background in the low mass range. Reports includes derivatization of the analyte molecules
to a higher molecular weight6 or using a matrix with higher molecular weight such as
porphyrin (MW 974.6).7 Extra sample preparation was then needed, thereby limiting the
classes of analytes that can be detected. It has been observed that matrix ions can be
suppressed dramatically and sometimes complete suppression can be achieved under well
controlled conditions.8, 9 For example, surfactant additives such as cetyltrimethylammonium
bromide (CTAB) have been reported to substantially suppress the background from α-cyano-
4-hydroxycinnamic acid (CHCA).10 Laser intensity and the relative molar ratio of matrix to
analyte are the major parameters to adjust. However a suitable molar ratio is not always
achievable especially for native biological samples.
Many inorganic materials have been tested as matrixes for surface-assisted laser
desorption/ionization (SALDI), including different metal powders and metal oxide
nanoparticles such as Ag, Au, Co, Al, Mn, Mo, Zn, Sn, W, Fe3O4, SnO2, TiO2, WO3, ZnO,
157
etc.11-15 Generally, those SALDI MS can provide a cleaner background than conventional
MALDI MS as no interference peaks from fragment ions of the organic matrixes were
present. Another matrix-free approach for laser desorption/ionization on porous silicon
(DIOS) was extensively studied since 1999.16, 17 Porous silicon surfaces were etched from
crystalline silicon chips with hydrofluoric acid and functionalized as the laser
desorption/ionization matrix as well as trapping agents for analyte molecules. Small
molecules including pharmaceuticals, nucleic acids, carbohydrates, and steroids were
successfully detected.18-20 In a more recent work, commercially available silicon
nanoparticles were utilized as an LDI matrix and the silicon powder preparation was
optimized for the analysis of small molecules.21 Different kinds of carbon materials,
including graphite particles,22 graphite plates,23, 24 graphite suspension in different solvents,25-
27 graphite trapped in silicone polymer,28 activated carbon powders,29 functionalized carbon
nanotubes30-33 and fullerenes,34 and more recently pencil lead,35-37 have been suggested as
alternative matrixes for LDI MS. Many kinds of analyte molecules over a wide mass range
(100-6000 Da) have been detected, such as peptides,20, 26-29, 31, 33-36 phospholipids,25
oligosaccharides,30-33, 35 fatty acids,24, 36 synthetic polymers7, 23, 26, 31, 32, 35, 37 and other various
organic compounds.7, 15, 22-29, 31, 33-37 A more detailed description of graphite-LDI can be
found in our previous paper.38 Recent reviews about small molecules MALDI MS39 and
matrix-free LDI MS can be found elsewhere.40
Imaging mass spectrometry (IMS) has proven to be a powerful technology for direct
profiling and imaging of elements and biomolecules in tissue sections. Secondary ion mass
spectrometry (SIMS),41, 42 MALDI43-47 and direct electrospray ionization (DESI)48 have been
applied as desorption/ionization techniques for the IMS of molecules such as metal elements,
peptides, proteins, lipids and other metabolites. SIMS has the best spatial resolution among
the three and DESI requires the least sample preparation and allows true in situ measurement
with the simplest instrumentation.48 The spatial resolution of MALDI IMS is in between
158
SIMS and DESI IMS, usually ranging from 80-200 µm in diameter. The diverse choices of
lasers and matrixes make MALDI MS suitable for fast, simultaneous and high-throughput
analyses of metabolites from tissue samples. MALDI equipped with UV laser has been
successfully demonstrated for the imaging of peptides, proteins and lipids.43-47 Due to the
high background problem as discussed earlier there is limited application of UV-MALDI for
imaging of small molecules (< 500 Da). Infrared (IR) MALDI was introduced recently as a
technique for imaging small metabolites from fruit samples.49 Water is used as the natural
matrix for IR MALDI, but it is inevitable that the sample may dry out during the process of
IR irradiation. Different locations will thus give different sensitivities due to inhomogeneous
water content. So far, molecules that can be detected by IR MALDI are quite limited, either
because of the low desorption/ionization efficiency or low detection sensitivity. Furthermore,
the spatial resolution of IR-IMS is inherently worse than that of UV-IMS.
Previously we demonstrated that colloidal graphite was a good LDI matrix for the
analysis of molecules in 500-1000 Da range, such as different lipid species.38 This matrix
contains fine particles and is spatially homogeneous, making it suitable for quantitative
imaging. The colloidal property also allows it to be easily sprayed to form a layer on top of
tissue samples and thus simplifies imaging experiments. In this study, we investigated the
applicability of colloidal graphite as an alternative LDI matrix for the analysis of even
smaller metabolite molecules. Fruits contain many kinds of small molecules such as long-
chain fatty acids, small oligosaccharides, and flavonoids; so they serve as good systems to
test the performance. GALDI MS and tandem MS were used to identify the ionized species,
while IMS and IMS/MS were utilized to map the distribution of those molecules in fruit
slices.
159
Experimental Section
Standards such as long-chain fatty acids, oligosaccharides and flavonoids were
purchased from Sigma-Aldrich (St. Louis, MO). Dihydroxybenzoic acid (DHB) from Bruker
Daltonics (Billerica, MA) and CHCA solution from Agilent Technologies (Palo Alto, CA)
were used as standard MALDI matrixes. 2-Propanol-based colloidal graphite aerosol spray
(Aerodag G) was obtained from Acheson Colloids (Port Huron, MI). Pure water was
obtained from a MilliQ water purification system (Billerica, MA). All other chemicals were
purchased from Fisher Scientific (Fairlawn, NJ).
Apple and strawberry fruits were purchased from a local grocery store. Apple skin
was peeled off by a sharp razor blade and attached to the stainless steel plate by double-sided
tape. Apple juice was collected from crushed flesh onto a glass slide, dried and used directly.
A cryostat from International Equipment Co. (Needham Heights, MA) was used for
cryosectioning. Fruit chunks were dipped in liquid N2 before they were cryosectioned into
about 15-µm thick specimens and stored at –20 °C before mass spectrometric analysis. No
optimum cutting temperature (OCT) compounds were used to embed the fruit samples as the
interference to mass spectra from OCT compounds is known.50 Sectioned fruit slices were
directly transferred and mounted onto the stainless-steel plate. Before applying colloidal
graphite solution, the slices were dried under moderate vacuum (~50 Torr) at room
temperature for half an hour.
Long-chain fatty acid standards were prepared by dissolving stearic acid (C18, MW
284.48), pentacosanoic acid (C25, MW 382.66), hexacosanoic acid (C26, MW 396.69),
octacosanoic acid(C28, MW 424.74), and melissic acid (C30, MW 452.80) in chloroform to
a final concentration of 200 pmole/µl each. For flavonoid standards, quercetin (MW 302.24),
kaempferol (MW 286.23), phloretin (MW 274.27), and apigenin (MW 270.24) were
dissolved individually in DMSO to give concentration of 5mg/ml of each; then the four
standard solution were mixed and further diluted to a final concentration of 200 pmole/µl
160
each in water/acetonitrile/trifluoroacetic acid (49.95/49.95/0.1). Oligosaccharide standards
were prepared by dissolving ribose (MW 150.13), glucose (MW 180.16), sucrose (MW
342.30), N-acetyl-D-lactosamine (LacNAc) (MW 383.35), maltotriose (MW 504.44), and
maltotetraose (MW 666.58) in water/acetonitrile/trifluoroacetic acid (49.95/49.95/0.1) to a
final concentration of 100 ng/µl each. 20 mg/mL DHB solution in 70% methanol and 30%
water (containing a 0.1% trifluoroacetic acid) was prepared. Commercial Agilent CHCA
solution at 6 mg/ml in 36/56/8 methanol/acetonitrile/water was purchased and used directly.
Four times dilution of colloidal graphite solution with 2-propanol was used for GALDI MS
and IMS.
For all mass spectrometric analysis and IMS, an LTQ linear ion trap mass
spectrometer equipped with vMALDI source (Thermo Electron, Mountain View, CA) was
used. The N2 laser (337 nm) is guided to the source by a fiber-optic cable and has a
maximum output of 280 µJ/pulse (before entering the optical fiber cable). The measured
laser spot size is ~100 µm in diameter on the sample plate surface. A more detailed
description of the LTQ with vMALDI source has been reported elsewhere.46
For conventional MALDI MS, 1 µL of DHB or CHCA matrix solution was applied
onto the stainless-steel sample plate and let to dry in air, followed by 1 µL of sample solution
on top of the matrix crystals. For GALDI MS of standards, 0.5 µL of diluted colloidal
graphite solution was applied onto the stainless-steel sample plate by a micropipette and let
to dry in air. Then 1 µL of standard solution was applied on top of the graphite spot. To
obtain mass spectra from apple juice with GALDI, 1 µl of fresh apple juice was applied onto
a dried (0.5 µl) graphite spot. For apple peels and fruit slices, diluted colloidal graphite
solution was applied by a double-action airbrush (Aztek A470 with a 0.30 mm nozzle from
Testor, Rockford, IL). The whole fruit slice was covered with colloidal graphite
homogeneously by spraying with 20 psi air pressure and 15 cm away from the sample plate
161
for 30 s. Peak identification was made by comparing both mass and tandem mass spectra
with those of standards.
Optical images of fruit slices were taken inside the vMALDI source before IMS.
Serial optical images were taken every 1 mm movement of the sample stage in either x- or y-
direction. Each segment of the images has a size of 140 pixels by 170 pixels. These segments
of optical images were reconstructed as one optical image for one fruit slice. To collect mass
spectra, the same sample plate was rastered with 100 µm steps. For each raster point, a mass
spectrum was recorded for desorbed ions and integrated over 3-5 laser shots. In the cases of
IMS/MS, target precursor ions (m/z 191 or m/z 301) were first selected based on the mass
spectral profiles of strawberry. Then the first-generation product-ion spectra of the selected
precursor ion were collected from all rastering points on the strawberry slice. More laser
shots were required for MS2 experiments and so 9 laser shots were averaged for each raster
point.
Custom software “vMALDI data browser” (Version 1.0) was used to extract mass
spectra from specific locations and generate chemically selective images. This software was
provided by the instrument vendor (Thermo Electron, Mountain View, CA ). The mass
window for generating images was 0.5 Da. Intensities of the selected ion were normalized by
dividing the total ion current of each mass spectrum. Then, chemically selective images were
plotted as 3-D maps with the 3rd dimension being the normalized intensity.
Results and Discussion
MALDI and GALDI MS of Standard Mixtures of Fatty Acids, Flavonoids and
Oligosaccharides. Different classes of compounds were selected to compare the
performance of conventional MALDI and GALDI. Long-chain fatty acids (C18-C30) were
selected as one group of standards because of their important roles in many metabolic
162
pathways of living organisms.51 Fatty acids can be detected by GC-MS with proper
derivatization such as esteritication to reduce the polarity and increase volatility.52 HPLC-ESI
MS can detect native fatty acids in the negative-ion mode through deprotonization.53
However, the basic pH condition which is required to produce such negative ions is not
compatible with most reverse-phase C18 columns that typically require acidic mobile phases.
To overcome this problem, judicious derivatization was needed.54 MALDI MS can eliminate
the separation step but conventional MALDI matrixes do not work well for detection due to a
high background and ion suppression in that mass region. This is underscored by the fact that
none of the selected standard fatty acids were detected with DHB or CHCA in either
positive-ion mode or negative-ion mode (data not shown). With GALDI, all five fatty acids
(100 ng each) were detected as deprotonated peaks ([M – H]–), as shown in Figure 1. It is
noteworthy that the background in the negative-ion mode of GALDI is very clean up to at
least m/z 1000 without any pretreatment. In fact, there are only a few low number carbon
cluster ions, such as C12–-C14
–; and the intensity of such peaks are much lower than those for
the analytes. GALDI is very sensitive to detect those fatty acids in negative ion mode. Fig. 2
shows the spectrum of the fatty acid mixture with sample loading of 100fmole of each on a
3mm-in-diameter spot. Such fatty acids can also be detected as potassium adduct ions ([M +
K]+) in positive ion mode (data not shown), though the detection sensitivity (detection limit:
50pmole/spot) is not as good as in negative ion mode .
Another group of standards tested were natural phenolic molecules, the flavonoids. It
was estimated that 2% of all carbon photosynthesized by plants is converted into flavonoids
or related compounds.55 They have been reported to have antioxidant, antiatherosclerotic and
anti-neurodegenerative properties, and are also known to be beneficial for the prevention of
chronic diseases like cancer and heart diseases.56, 57 In the positive-ion mode, quercetin,
kaempferol and apigenin were detected as [M + H]+ in both MALDI and GALDI
experiments (Fig. 3a-c). [M + Na]+ and [M + 2Na-H]+ ions for these three flavonoids can also
163
be detected with GALDI. Phloretin was not detected with any of the three matrixes. This
suggests that the center ring in the other three flavonoids plays an important role for
protonation. However, it was possible to detect phloretin with GALDI in the negative-ion
mode (Fig. 3f). Several peaks in the region of m/z 150-200 in the spectra were identified as
in-source-fragments for the flavonoids by using tandem MS. For example, m/z 167 was
identified as a fragment of phloretin while m/z 151 and m/z 179 were fragments of quercetin.
Unlike GALDI, which has a clean background, with DHB and CHCA, matrix peaks are
predominant and none of the four flavonoid standards were detected in the negative-ion
mode (Fig. 3d-e).With GALDI, negative ion mode provides better sensitivity for detection of
those flavonoid standards than positive ion mode does, and the detection limit are 50 fmole
and 200 fmole/3mm-in-diameter spot, respectively.
The third group of standards selected were oligosaccharides. Due to the lack of acidic
or basic groups, they are difficult to be ionized with conventional MALDI. DHB can be used
to detect large oligosaccharides (> 1000 Da)58 but smaller oligosaccharides are strongly
interfered with by matrix ions. We tested oligosaccharides between 150-700 Da with DHB
(Fig. 4a), CHCA (Fig. 4b), and graphite matrixes (Fig. 5). With CHCA, only larger
molecules like LacNAc, maltotriose and maltotetraose can be detected as [M + Na]+ but not
the smaller ones such as ribose, glucose and sucrose. DHB works slightly better than CHCA
in that the small oligosaccharides such as glucose and sucrose can be detected. However, the
intensity of the small saccharide peaks was very low. For example [ribose + Na]+ and
[glucose + Na]+ peaks are strongly suppressed as they are very close to matrix peaks [DHB +
Na]+ and [DHB + 2Na – H]+ respectively. LacNAc has the highest detection sensitivity
among all the oligosaccharides. This may suggest that the N-acetyl group has more affinity
for sodium ions than the other moieties of oligosaccharides. In both the CHCA and DHB
MALDI mass spectra (Fig. 4a-b), the matrix peaks were predominant and oligosaccharides
peaks were strongly interfered with. With GALDI, all six oligosaccharides were detected as
164
[M + Na]+ with decent signal-to-noise ratios, as shown in Fig. 5. [M + Na -18]+ peaks were
also observed for all five oligosaccharides as water loss (Fig. 5). Neutral loss (NL) of 60 are
known to result from the major cross-ring cleavage fragments of oligosaccharides with 1,4
linkages.59 Here, m/z 629 and m/z 467 were observed as such fragments of maltotetraose and
maltotriose respectively. The peak at m/z 305 corresponds to the major fragment of LacNAc.
All of the above fragments were verified by tandem MS (data not shown). All cross-ring
cleavage fragments were marked with asterisks in Fig. 5. That the positive-ion mode of
GALDI gave a noisier background compared to the negative-ion mode did not prevent the
successful detection of small oligosaccharides. Small oligosaccharides such as hexose
(glucose or fructose) and sucrose can also be detected as [M – H]– under the negative-ion
mode peaks, vide infra. The detection sensitivity of oligosaccharides with GALDI under
positive ion mode is better than with GALDI under negative ion mode. For instance, the
detection limit of sucrose is 20pmole and 100pmole/3mm-in-diameter spot, respectively.
This can be understood as oligosaccharides have neither carboxyl group as fatty acids, or
aromatic rings as flavonoids which can stabilize the deprotonated phenol group.
Direct Detection of Small Metabolite Molecules from Fruit Samples. Because of
the sensitivity and background issues, GALDI in the positive-ion mode was used to detect
oligosaccharides and the negative-ion mode was used to detect other metabolite molecules
from fruit samples. Unlike experiments with the standards, colloidal graphite solution was
applied on top of the sample, otherwise the graphite particles may not be accessed by laser
irradiation. For imaging purposes, homogeneous coverage over the sample area is a must.
The application methodology has been optimized for IMS of mouse brain tissues in our
previous work38 and was used in this study without modification.
A list of small molecules detected directly from apple and strawberry samples were
summarized in Table I, with concentration data wherever available.60-65 All identifications
were made by comparing the MS spectra with those of standards. MS/MS data were also
165
compared with those of standards except those at too low a concentration to give meaningful
MS/MS data, as marked with asterisks. Tandem MS data is indispensable to identify isobaric
ions, such as m/z 191, which were observed from both apple and strawberry. Tandem mass
data of m/z 191 from apple and strawberry flesh are shown in Fig. 6. Notably, citric acid gave
a predominant product ion at m/z 111 while quinic acid has specific product ions at m/z 85,
93 and 127, as reported previously.66 By tandem MS it was confirmed that the peaks at m/z
191 were from quinic acid in apple while such peak came from citric acid in strawberry.
Fig. 7 shows typical mass spectra taken from different parts of the apple. Fruits in
supermarkets were always coated with a thin layer of wax for preservation and for better
appearance. Fatty acids are one major class of components of wax and many of those up to
C28 fatty acids were detected from apple peel, as shown in Fig. 7a. Other compounds such as
sugars or flavonoids were absent from the mass spectrum. The reason may be that such
compounds were covered by the wax layer and were not accessible. Fatty acids may be
naturally present on the apple peels as well; however, those cannot be discriminated from the
species in the artificial wax layer.
As shown in Fig. 7b, fresh apple juice gave malic acid, quinic acid, palmitic acid, and
linolenic acid in the negative-ion mode. Hexose (glucose or fructose) and sucrose were
detected as deprotonated ions as m/z 179 and m/z 341, respectively. Quercetin is one of the
major flavonoids contained in apple and it was also detected. Fig. 7c shows the spectrum in
the positive-ion mode. Sodium adduct ions of hexose and sucrose were detected, as well as
potassium adduct ions. According to the USDA nutrient database, fruits usually contain
much higher amounts of potassium than sodium (90 mg vs. 0 mg/100 g for apple and 292 mg
vs. 37 mg/100 g for strawberry).64
Fig. 7d is the representative spectrum taken from the apple core (endocarp). Here,
organic acids such as malic acid and quinic acid were still observed, but the relative
intensities were not as high as those from juice. On the other hand, higher amounts of
166
flavonoids such as phloretin and quercetin accumulated in the endocarp region. Apple peel
also contains a lot of flavonoids and quercetin was the predominant one found on the inside
of apple peel (data not shown).
Fig. 8a and 8b are representative mass spectra taken from strawberry. As reported
previously with IR-MALDI,49 strawberry contains a lot of citric acid. In our study, citric acid
(m/z 191) was always the most intense peak in the negative-ion mode. GALDI can also detect
those compounds which were too low in concentration to be detected by IR-MALDI, such as
ascorbic acid and ellagic acid. Quercetin and kaempferol was also detected in the red part of
strawberry (more details in IMS data below). Other deprotonated ions include palmitic acid,
oleic acid, apigenin, hexose, and sucrose, as marked in Fig. 8a. Similar to the previous
report,49 citric acid and sucrose were barely detected in the seed region, as shown in Fig. 8b.
Instead, C16 and unsaturated C18 fatty acids were the major components.
IMS of Metabolites from Apple and Strawberry. IMS in the negative-ion mode
was performed to show the detailed distributions of different metabolite molecules on apple
and strawberry slices, as Fig. 9 and Fig. 10 respectively. Small molecules such as malic acid,
quinic acid, and sucrose distributed relatively evenly over the apple flesh part, as shown in
Fig. 9b, c, and e. Long-chain fatty acid such as linoleic acid was detected all over the slice
but more accumulated along the core line and bundles, as shown in Fig. 9d. Flavonoids also
accumulated more in the core region but not in the flesh, as shown in Fig. 9f-i. Unlike
linoleic acid, they seem to be only enriched in the bundles (both sepal and pedal), but not
along the core line. Another interesting feature is that flavonoids were found in the ventral
carpellary bundle (center of the apple slice) except for quercetin.
The images shown in Fig. 10 were scanned from a pie-shaped strawberry slice, with
the red skin on the right-hand side and one seed in the middle-right part. Citric acid, apigenin,
hexose and sucrose were distributed all over the flesh, while fatty acids such as linolenic and
linoleic acids accumulated on the seed. The peak at m/z 301.2 was detected all over the
167
strawberry with a relatively constant concentration in the flesh, but showed higher local
intensity on the outside (red part) and the seed areas. The candidates were quercetin (MW
302.24) and ellagic acid (MW 302.20) and the two species could be distinguished in tandem
MS. Fig. 11a and 6b show the MS/MS spectra of ellagic acid and quercetin standards. m/z
151 and m/z 179 were specific fragments of quercetin due to the cleavage of the center ring,
as reported previously;56, 67 while ellagic acid is more rigid and only small fragments such as
NL 28 and NL 44 were observed. MS/MS product ion spectra from strawberry flesh were
similar to those from ellagic acid (Fig. 11c), while those at m/z 301.2 on the edge gave
quercetin-like fragments (Fig. 11d). The chemically selective ion image at m/z 179 with
precursor ions at m/z 301.2 is shown in Fig. 11e. The ambiguity in Fig. 10i is thus resolved.
Similarly, both citric acid (MW 192.13) and quinic acid (MW 192.17) were detected
as m/z 191 with GALDI in the negative-ion mode. IMS/MS (data not shown) with precursor
ions at m/z 191.2 showed that product ions at m/z 111.3 were detected all over the strawberry
slice, while no specific fragments of quinic acid (m/z 85 or m/z 93) were observed. This
suggests that all m/z 191 ions from strawberry were from citric acid (Fig. 10c).
Acknowledgements
E.S.Y. thanks the Robert Allen Wright Endowment for Excellence for support. The
Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University
under Contract No. DE-AC02-07CH11358. This work was supported by the Director of
Science, Office of Basic Energy Sciences, Divisions of Chemical Sciences and Biosciences.
168
Table I. List of compounds detected directly from apple and strawberry and their average
concentrations from the literature. All species have been checked by m/z against standards,
and all but those with marked with asterisks have been checked by tandem MS.
Apple Strawberry
m/z Peaks assigned Concentrations g/Kg
m/z Peaks assigned Concentrations g/Kg
Organic acids
133 [M-H]– Malic Acid 4.9 *10-3 60 175 [M-H]– Ascorbic Acid* 0.59
191 [M-H]– Quinic Acid 0.76 *10-3 60 191 [M-H]– Citric Acid 6.9-12.6 62
255 [M-H]– Palmitic Acid 0.24 255 [M-H]– Palmitic Acid 0.12
277 [M-H]– Linolenic Acid 0.09 277 [M-H]– Linolenic Acid 0.65
279 [M-H]– Linoleic Acid 0.43 279 [M-H]– Linoleic Acid 0.90
281 [M-H]– Oleic Acid 0.07 281 [M-H]– Oleic Acid 0.42
Phenolics
273 [M-H]– Phloretin 4.7 *10-3 60 269 [M-H]– Apigenin 0.00-0.01
289 [M-H]– Epicatechin 47.1 *10-3 285 [M-H]– Kaempferol 4.6 *10-3
301 [M-H]– Quercetin 45.7 *10-3 301 [M-H]– Ellagic Acid 0.06-0.5 63
435 [M-H]– Phloridzin* 55.9 *10-3 61 301 [M-H]– Quercetin 11.4 *10-3
447 [M-H]– Quercetin Glucosides*
0.13 61 431 [M-H]– Apigenin Glucosides*
Oligo- saccharides
179 [M-H]–
203 [M+Na]+
219 [M+K]+
Glucose/ Fructose*
83.3 179 [M-H]–
203 [M+Na]+ 219 [M+K]+
Glucose/ Fructose*
44.3
341 [M-H]–
365 [M+Na]+
381 [M+K]+
Sucrose 20.7 341 [M-H]–
365 [M+Na]+ 381 [M+K]+
Sucrose
4.7
Except noted, all other concentration data come from (1) See the USDA National Nutrient
Database at http://www.nal.usda.gov/fnic/foodcomp/search/, and (2) See the USDA National
Nutrient Database at http://www.nal.usda.gov/fnic/foodcomp/Data/Flav/flav.pdf
169
References
(1) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301.
(2) Bahr, U.; Deppe, A.; Karas, M.; Hillenkamp, F.; Giessmann, U. Anal. Chem. 1992, 64,
2866-2869.
(3) Kaufmann, R.; Kirsch, D.; Spengler, B. Int. J. Mass Spectrom. Ion Processes 1994,
131, 355-385.
(4) Lecchi, P.; Pannell, L. K. J. Am. Soc. Mass. Spectrom. 1995, 6, 972-975.
(5) Finke, B.; Stahl, B.; Pfenninger, A.; Karas, M.; Daniel, H.; Sawatzki, G. Anal. Chem.
1999, 71, 3755-3762.
(6) Tholey, A.; Wittmann, C.; Kang, M. J.; Bungert, D.; Hollemeyer, K.; Heinzle, E. J.
Mass Spectrom. 2002, 37, 963-973.
(7) Ayorinde, F. O.; Hambright, P.; Porter, T. N.; Keith, Q. L. Rapid Commun. Mass
Spectrom. 1999, 13, 2474-2479.
(8) Knochenmuss, R.; Dubois, F.; Dale, M. J.; Zenobi, R. Rapid Commun. Mass
Spectrom. 1996, 10, 871-877.
(9) Knochenmuss, R.; Karbach, V.; Wiesli, U.; Breuker, K.; Zenobi, R. Rapid Commun.
Mass Spectrom. 1998, 12, 529-534.
(10) Guo, Z.; Zhang, Q. C.; Zou, H. F.; Guo, B. C.; Ni, J. Y. Anal. Chem. 2002, 74, 1637-
1641.
(11) Sluszny, C.; Yeung, E. S.; Nikolau, B. J. J. Am. Soc. Mass. Spectrom. 2005, 16, 107-
115.
(12) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T.; Matsuo, T. Rapid
Commun. Mass Spectrom. 1988, 2, 151-153.
(13) McLean, J. A.; Stumpo, K. A.; Russell, D. H. J. Am. Chem. Soc. 2005, 127, 5304-
5305.
170
(14) Kinumi, T.; Saisu, T.; Takayama, M.; Niwa, H. J. Mass Spectrom. 2000, 35, 417-422.
(15) Chen, C. T.; Chen, Y. C. Anal. Chem. 2005, 77, 5912-5919.
(16) Wei, J.; Buriak, J. M.; Siuzdak, G. Nature 1999, 399, 243-246.
(17) Shen, Z. X.; Thomas, J. J.; Averbuj, C.; Broo, K. M.; Engelhard, M.; Crowell, J. E.;
Finn, M. G.; Siuzdak, G. Anal. Chem. 2001, 73, 612-619.
(18) Pihlainen, K.; Grigoras, K.; Franssila, S.; Ketola, R.; Kotiaho, T.; Kostiainen, R. J.
Mass Spectrom. 2005, 40, 539-545.
(19) Compton, B. J.; Siuzdak, G. Spectroscopy-an International Journal 2003, 17, 699-
713.
(20) Li, Q.; Ricardo, A.; Benner, S. A.; Winefordner, J. D.; Powell, D. H. Anal. Chem.
2005, 77, 4503-4508.
(21) Wen, X. J.; Dagan, S.; Wysocki, V. H. Anal. Chem. 2007, 79, 434-444.
(22) Zumbuhl, S.; Knochenmuss, R.; Wulfert, S.; Dubois, F.; Dale, M. J.; Zenobi, R. Anal.
Chem. 1998, 70, 707-715.
(23) Kim, H. J.; Lee, J. K.; Park, S. J.; Ro, H. W.; Yoo, D. Y.; Yoon, D. Y. Anal. Chem.
2000, 72, 5673-5678.
(24) Park, K. H.; Kim, H. J. Rapid Commun. Mass Spectrom. 2001, 15, 1494-1499.
(25) Peng, S.; Edler, M.; Ahlmann, N.; Hoffmann, T.; Franzke, J. Rapid Commun. Mass
Spectrom. 2005, 19, 2789-2793.
(26) Dale, M. J.; Knochenmuss, R.; Zenobi, R. Anal. Chem. 1996, 68, 3321-3329.
(27) Sunner, J.; Dratz, E.; Chen, Y. C. Anal. Chem. 1995, 67, 4335-4342.
(28) Li, X. P.; Wilm, M.; Franz, T. Proteomics 2005, 5, 1460-1471.
(29) Chen, Y. C.; Shiea, J.; Sunner, J. J. Chromatogr. A 1998, 826, 77-86.
171
(30) Pan, C. S.; Xu, S. Y.; Hu, L. G.; Su, X. Y.; Ou, J. J.; Zou, H. F.; Guo, Z.; Zhang, Y.;
Guo, B. C. J. Am. Soc. Mass. Spectrom. 2005, 16, 883-892.
(31) Ren, S. F.; Guo, Y. L. Rapid Commun. Mass Spectrom. 2005, 19, 255-260.
(32) Ren, S. F.; Zhang, L.; Cheng, Z. H.; Guo, Y. L. J. Am. Soc. Mass. Spectrom. 2005, 16,
333-339.
(33) Xu, S. Y.; Li, Y. F.; Zou, H. F.; Qiu, J. S.; Guo, Z.; Guo, B. C. Anal. Chem. 2003, 75,
6191-6195.
(34) Ugarov, M. V.; Egan, T.; Khabashesku, D. V.; Schultz, J. A.; Peng, H. Q.;
Khabashesku, V. N.; Furutani, H.; Prather, K. S.; Wang, H. W. J.; Jackson, S. N.;
Woods, A. S. Anal. Chem. 2004, 76, 6734-6742.
(35) Black, C.; Poile, C.; Langley, J.; Herniman, J. Rapid Commun. Mass Spectrom. 2006,
20, 1053-1060.
(36) Langley, G. J.; Herniman, J. M.; Townell, M. S. Rapid Commun. Mass Spectrom.
2007, 21, 180-190.
(37) Berger-Nicoletti, E.; Wurm, F.; Kilbinger, A. F. M.; Frey, H. Macromolecules 2007,
40, 746-751.
(38) Cha, S.; Yeung, E. S. Anal. Chem. 2007, 79, 2373-2385.
(39) Cohen, L. H.; Gusev, A. I. Anal. Bioanal. Chem. 2002, 373, 571-586.
(40) Peterson, D. S. Mass Spectrom. Rev. 2007, 26, 19-34.
(41) Altelaar, A. F. M.; Klinkert, I.; Jalink, K.; de Lange, R. P. J.; Adan, R. A. H.; Heeren,
R. M. A.; Piersma, S. R. Anal. Chem. 2006, 78, 734-742.
(42) Ostrowski, S. G.; Van Bell, C. T.; Winograd, N.; Ewing, A. G. Science 2004, 305, 71-
73.
(43) Caprioli, R. M.; Farmer, T. B.; Gile, J. Anal. Chem. 1997, 69, 4751-4760.
(44) Chaurand, P.; Caprioli, R. M. Electrophoresis 2002, 23, 3125-3135.
172
(45) Chaurand, P.; Cornett, D. S.; Caprioli, R. M. Curr. Opin. Biotechnol. 2006, 17, 431-
436.
(46) Garrett, T. J.; Prieto-Conaway, M. C.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford, G.;
Yost, R. A. Int. J. Mass Spectrom. 2007, 260, 166-176.
(47) Garrett, T. J.; Yost, R. A. Anal. Chem. 2006, 78, 2465-2469.
(48) Wiseman, J. M.; Ifa, D. R.; Song, Q. Y.; Cooks, R. G. Angew. Chem. Int. Ed. 2006,
45, 7188-7192.
(49) Li, Y.; Shrestha, B.; Vertes, A. Anal. Chem. 2007, 79, 523-532.
(50) Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. J. Mass Spectrom. 2003, 38, 699-708.
(51) Claudio, G. A., P.; Simopoulos, E. T. Effects of fatty acids and lipids in health and
disease; Basel: New York 1994.
(52) Halket, J. M.; Zaikin, V. G. Eur. J. Mass Spec. 2004, 10, 1-19.
(53) Nagy, K.; Jakab, A.; Fekete, J.; Vekey, K. Anal. Chem. 2004, 76, 1935-1941.
(54) Yang, W. C.; Adamec, J.; Regnier, F. E. Anal. Chem. 2007, ASAP.
(55) Robards, K.; Antolovich, M. Analyst 1997, 122, R11-R34.
(56) Seeram, N. P.; Lee, R.; Scheuller, H. S.; Heber, D. Food Chem. 2006, 97, 1-11.
(57) Törrönen, R. a. M., K. Acta Hort. (ISHS) 2002, 797-803.
(58) Harvey, D. J.; Martin, R. L.; Jackson, K. A.; Sutton, C. W. Rapid Commun. Mass
Spectrom. 2004, 18, 2997-3007.
(59) Asam, M. R.; Glish, G. L. J. Am. Soc. Mass. Spectrom. 1997, 8, 987-995.
(60) Gokmen, V.; Artik, N.; Acar, J.; Kahraman, N.; Poyrazoglu, E. Eur. Food
Res.Technol. 2001, 213, 194-199.
(61) Lee, K. W.; Kim, Y. J.; Kim, D. O.; Lee, H. J.; Lee, C. Y. J. Agric. Food. Chem. 2003,
51, 6516-6520.
173
(62) Skupien, K.; Oszmianski, J. Eur. Food Res.Technol. 2004, 219, 66-70.
(63) Williner, M. R.; Pirovani, M. E.; Guemes, D. R. Journal of the Science of Food and
Agriculture 2003, 83, 842-845.
(64) See the USDA National Nutrient Database at
http://www.nal.usda.gov/fnic/foodcomp/search/ Accessed on Mar. 2007.
(65) See the USDA National Nutrient Database at
http://www.nal.usda.gov/fnic/foodcomp/Data/Flav/flav.pdf Accessed on Mar. 2007.
(66) Ng, L. K.; Lafontaine, P.; Vanier, M. J. Agric. Food. Chem. 2004, 52, 7251-7257.
(67) Chen, M. L.; Song, F. R.; Guo, M. Q.; Liu, Z. Q.; Liu, S. Y. Rapid Commun. Mass
Spectrom. 2002, 16, 264-271.
174
Figure Captions
Figure 1. Mass spectrum of fatty acid standards (C18-C30 fatty acids) with GALDI in
the negative-ion mode. Sample loading: 200 pmole of each.
Figure 2. Mass spectrum and detection limit of fatty acid standards (C18-C30 fatty
acids) with GALDI in the negative-ion mode. Sample loading: 100 fmole of
each.
Figure 3. Comparison of MALDI and GALDI spectra of flavonoid standards: quercetin
(MW 302.24), kaempferol (MW 286.23), phloretin (MW 274.27) and
apigenin (MW 270.24); Sample loading: 200 pmole of each. Apigenin,
kaempferol and quercetin were observed as [M + H]+ with DHB (a) and
CHCA (b) in the positive ion mode. With GALDI in the positive-ion mode (c),
apigenin, kaempferol and quercetin are detected as [M + H] + (m/z 271, m/z
286, and m/z 303, respectively), [M+Na] + (m/z 293, m/z 309, and m/z 325,
respectively), and [M + 2Na - H] + (m/z 315, m/z 331, and m/z 347,
respectively). Phloretin was not observed either. None of the four flavonoids
can be detected with DHB (d) or CHCA (e) in the negative ion mode;
However, all the four flavonoids are detected as [M - H]- ions with GALDI in
the negative-ion mode (f). Fragment from phloretin (m/z 167) and fragments
from quercetin (m/z 151 and m/z 179) are observed in the lower mass range as
well.
Figure 4. Mass spectrum of oligosaccharide standards with DHB (a) and CHCA (b) in
the positive-ion mode. Sample loading: 100 ng of each.
175
Figure 5. Mass spectrum of oligosaccharide standards with GALDI in the positive-ion
mode. Sample loading: 100 ng of each. Peaks with ●: water loss fragments;
peaks with *: major ring-cleavage fragments.
Figure 6. Product ion spectra of m/z 191 from (a) quinic acid standard; (b) apple flesh;
(c) citric acid standard; and (d) strawberry flesh. All spectra were collected in
the negative-ion mode.
Figure 7. Representative GALDI-MS spectra from different parts of apple. (a) fatty
acids composition on the outside of apple peel, negative-ion mode; (b) fresh
apple juice, negative-ion mode; (c) hexose and sucrose from fresh apple juice,
positive-ion mode; and (d) apple core, negative-ion mode. See Table I for
peak identification in (b) and (d).
Figure 8. Representative GALDI-MS spectra (negative-ion mode) from strawberry (a)
flesh; and (b) seed.
Figure 9. Chemically selective images of major ionic species identified from apple
endocarp region with GALDI in the negative-ion mode. (a) optical image
taken with reversed color; (b) malic acid; (c) quinic acid; (d) linoleic acid; (e)
sucrose; (f) phloretin; (g). epicatechin; (h) quercetin; and (i) phloridzin. All
peak intensities were normalized by dividing by the total ion current (TIC) of
each spectrum.
Figure 10. Chemically selective images of the major ionic species identified from
strawberry with GALDI in the negative-ion mode. (a) optical image taken
176
with reversed color. One seed was present as the darkest dot on the right-hand
side; (b) ascorbic acid; (c) citric acid; (d) linoleic acid; (e) hexose; (f) sucrose;
(g) apigenin; (h) kaempferol; and (i) m/z 301-301.5 ellagic acid + quercetin.
All peak intensities were normalized by dividing by the total ion current of
each spectrum.
Figure 11. Product ion spectra of m/z 301 from (a) ellagic acid standard; (b) strawberry
flesh; (c) quercetin standard; (d) edge of strawberry; and (e) chemically
selective image for product ion at m/z 179 (precursor ion m/z 301) of
strawberry slice with GALDI. All spectra were collected in the negative-ion
mode. The proposed fragment pathways are shown as inserts.67
177
100 150 200 250 300 350 400 450 5000
10
20
30
40
50
60
70
80
90
100
m/z
C18 Fatty Acid
C10 C13
C12C11
Rela
tive
Abu
ndan
ce (%
)
C25 Fatty Acid
C26 Fatty Acid
C28 Fatty Acid
C30 Fatty Acid
100 150 200 250 300 350 400 450 5000
10
20
30
40
50
60
70
80
90
100
m/z
C18 Fatty Acid
C10 C13
C12C11
Rela
tive
Abu
ndan
ce (%
)
C25 Fatty Acid
C26 Fatty Acid
C28 Fatty Acid
C30 Fatty Acid
Figure 1.
178
260 280 300 320 340 360 380 400 420 440 460 480 500m/z
0
10
20
30
40
50
60
70
80
90
100 283.33
395.50 423.50381.42
451.42
C25 Fatty Acid
C18 Fatty Acid
C26 Fatty Acid
C28 Fatty Acid
C30 Fatty AcidR
elat
ive
Abun
danc
e (%
)
Figure 2.
179
OHO
OH
OH
O
Apigenin
OHO
OH
OH
OH
O
Quercetin
OHO
OH
OH
OH
O
Kaempferol
OHHO
OH
OH
O
Phloretin
m/z250 290 310 380
271.42
310 320 330m/z
150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 340 350
Rel
ativ
e A
bund
ance
361.33
287.42
339.33303.42
271.50
273.42
0
10
20
30
40
50
60
70
80
90
100
317.42
[Apigenin + H]+
[Kaempferol + H]+
[Quercetin + H]+
a.
[Kaempferol + H]+
[Quercetin + H]+
[Apigenin + H]+
287.42 303.42
379.33
271.50b.
0
10
20
30
40
50
60
70
80
90
100
+
287.33 293.33
309.33325.25 331.33
347.25
303.33
315.33
[Apigenin + H]+
[Kaempferol + H]+[Quercetin + H]+
c.
0
10
20
30
40
50
60
70
80
90
100
260 270 280 300 320 330 340 350 360 370 390 400
[Quercetin - H]-f.
188.50e.
0
10
20
30
40
50
60
70
80
90
100
153.50
315.25d.
0
10
20
30
40
50
60
70
80
90
100
[Kaempferol - H]-
[Apigenin - H]-
151.33179.33
167.33
[Phloretin - H]-
269.25
273.25
269.25
301.17
0
10
20
30
40
50
60
70
80
90
100
OHO
OH
OH
O
Apigenin
OHO
OH
OH
OH
O
Quercetin
OHO
OH
OH
OH
O
Quercetin
OHO
OH
OH
OH
O
Kaempferol
OHO
OH
OH
OH
O
Kaempferol
OHHO
OH
OH
O
Phloretin
OHHO
OH
OH
O
Phloretin
m/z250 290 310 380
271.42
310 320 330m/z
150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 340 350150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 340 350
Rel
ativ
e A
bund
ance
361.33
287.42
339.33303.42
271.50
273.42
0
10
20
30
40
50
60
70
80
90
100
317.42
[Apigenin + H]+
[Kaempferol + H]+
[Quercetin + H]+
a.
361.33
287.42
339.33303.42
271.50
273.42
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
317.42
[Apigenin + H]+
[Kaempferol + H]+
[Quercetin + H]+
a.
[Kaempferol + H]+
[Quercetin + H]+
[Apigenin + H]+
287.42 303.42
379.33
271.50b.
0
10
20
30
40
50
60
70
80
90
100
[Kaempferol + H]+
[Quercetin + H]+
[Apigenin + H]+
287.42 303.42
379.33
271.50b.
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
+
287.33 293.33
309.33325.25 331.33
347.25
303.33
315.33
[Apigenin + H]+
[Kaempferol + H]+[Quercetin + H]+
c.
0
10
20
30
40
50
60
70
80
90
100
+
287.33 293.33
309.33325.25 331.33
347.25
303.33
315.33
[Apigenin + H]+
[Kaempferol + H]+[Quercetin + H]+
c.
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
260 270 280 300 320 330 340 350 360 370 390 400260 270 280 300 320 330 340 350 360 370 390 400
[Quercetin - H]-f.
188.50e.
0
10
20
30
40
50
60
70
80
90
100188.50e.
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
153.50
315.25d.
0
10
20
30
40
50
60
70
80
90
100
153.50
315.25d.
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
[Kaempferol - H]-
[Apigenin - H]-
151.33179.33
167.33
[Phloretin - H]-
269.25
273.25
269.25
301.17
0
10
20
30
40
50
60
70
80
90
100
[Kaempferol - H]-
[Apigenin - H]-
151.33179.33
167.33
[Phloretin - H]-
269.25
273.25
269.25
301.17
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
Figure 3.
180
100 200 300 400 500 600m/z
LacNAC406.25 Maltotetraose
689.25
x2
0
10
20
30
40
50
60
70
80
90
100
Maltotriose527.42
Sucrose365.25
Glucose203.33
Maltotetraose689.25
Maltotriose527.42
LacNAC406.25
a.
b.
Rel
ativ
e A
bund
ance
0
10
20
30
40
50
60
70
80
90
100
100 200 300 400 500 600m/z
LacNAC406.25 Maltotetraose
689.25
x2
0
10
20
30
40
50
60
70
80
90
100
Maltotriose527.42
Sucrose365.25
Glucose203.33
Maltotetraose689.25
Maltotriose527.42
LacNAC406.25
a.
b.
Rel
ativ
e A
bund
ance
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
Figure 4.
181
100
200
300
400
500
600
700
0102030405060708090100
m/z
Relative Abundance (%)
Rib
ose
Mal
totri
ose
LacN
Ac
Suc
rose
Glu
cose
Mal
tote
traos
e●
●●
●
●*
*
**
**
●
Figure 5.
182
5010
015
020
00102030405060708090100
m/z
Relative Abundance (%)
173.
25
171.
3315
5.33
147.
25
127.
33
111.
50
109.
33
93.3
3
85.4
2
5010
015
020
00102030405060708090100
m/z
Relative Abundance (%)
59.5
8
127.
33
111.
42
109.
42
93.5
0
85.5
0
71.5
0
171.
33
155.
2514
3.25
173.
25a. b.
5010
015
020
00102030405060708090100
m/z
Relative Abundance (%)
191.
25
5010
015
020
00102030405060708090100
m/z
Relative Abundance (%)
c. d.
111.
33
173.
08
163.
4214
7.25
129.
58
111.
33x5
173.
25
163.
2514
7.25
129.
42
191.
25
Figure 6.
183
100 150 200 250 300 350 400 450 5000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
100 150 200 250 300 350 400 450 5000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
100 150 200 250 300 350 400 450 5000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
100 150 200 250 300 350 400 450 5000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
a. Apple Peel (-)
d. Apple Endocarp (-)
c. Apple Juice (+)
b. Apple Juice (-)
C16
C18:3
C18:2
C18:1
C20 C22 C24 C26 C28
447.3341.5
301.2277.3
255.2
191.42
179.3
[Hexose + Na] +
[Hexose + K] +
[Sucrose + Na] +
[Sucrose + K] +
133.2
179.2
191.2
255.3
273.2
279.3
281.3
289.3
301.2341.2
435.2447.2
Figure 7.
184
100 150 200 250 300 350 4000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
AscobicAcid
Citric Acid
EllagicAcid
C18:1Fatty Acid
C16Fatty Acid
SucroseApigenin
a. Strawberry Flesh
100 150 200 250 300 350 4000
10
20
30
40
50
60
70
80
90
100
m/z
Rel
ativ
e A
bund
ance
(%)
C16Fatty Acid
EllagicAcid
C18:1Fatty Acid
C18:3Fatty Acid
C18:2Fatty Acid
b. Strawberry Seed
Figure 8.
185
Figure 9.
186
Figure 10.
187
Figure 11.
188
APPENDIX 3. SUPPORTING INFORMATION FOR CHAPTER 3
Sangwon Cha, Hui Zhang, Hilal I. Ilarslan, Eve Syrkin Wurtele, Libuse Brachova,
Basil J. Nikolau and Edward S. Yeung
Enhancing Spatial Resolution in Mass Spectral Imaging by ‘Oversampling’
Because 100 µm or 150 µm step size was too big to distinguish flower components—
carpels, petals, sepals and so on—and to recognize the boundaries of the flower components,
50 µm spacing was used for flower scanning. Scanning with a spacing which is smaller than
laser spot size is called oversampling. This oversampling can enhance the spatial resolution
of the chemical selective image without optical adjustments if complete ablation occurs in
each scanning spot (Jurchen et al., 2005). If complete ablation is not achieved, oversampling
just increases the number of pixels of the processed image but could not provide real
spatially-resolved chemical distribution information. The number of laser shots needed for
complete ablation was investigated experimentally by counting to when mass spectrum
started showing no or negligible signal. However, as Jurchen and coworkers pointed out
(Jurchen et al., 2005), the qualities of mass spectra collected near complete ablation point
were poor and this leads to poor sensitivities if all mass spectra collected at one point were
averaged.
To overcome this issue, ‘selective spectra averaging’ method with complete ablation
for one scanning point and ‘whole spectra averaging’ method without complete ablation for
one scanning point were performed. In selective spectra averaging, a series of mass spectra
were collected for the small area (9-16 rastering points on the sample) at one x, y coordinate
189
on the sample with a fixed number of laser shots per mass spectrum (10 laser shots in this
case) until no signal or negligible signal was observed. With an averaged number of
microscans (one microscan corresponds to getting one mass spectrum from the sample) until
complete ablation, data were collected over the sample surface. Intensity information with x,
y-coordinates for interesting m/z ranges were extracted by using the custom software from
Thermo (vMALDI Data Extract). This data were then separated to individual microscan
spectra. Under our experimental condition, 7 to 8 microscans were needed to achieve
complete ablation in most cases. For Arabidopsis flowers, averaged intensity values from the
first three to five microscans were used to generate chemically-selective images of flavonoids
and less than three microscans were averaged for the other classes of compounds. Images
from selective spectra averaging were generated by using in-house written Matlab
(Mathworks, Natick, MA, USA) program.
By using these two methods, there was no significant difference in terms of qualities
of chemical images for flavonoids but the two methods have advantages and disadvantages
over each other. Because of the need for complete ablation of graphite materials, ‘selective
spectra averaging’ method usually requires 1.3 to 2 times longer collection time compared to
‘whole spectra averaging’ method. ‘Whole spectra averaging’ method can reduce the data
collection time but there is a possibility of resulting in poor image quality if the sample
contains many different surface properties that can lead to different ion yields, or if
interesting molecules have very different ionization characteristics or different abundances.
These possible problems can be resolved in ‘selective spectra averaging’ method by varying
number of microscans to be averaged according to different ion species. For example, the ion
species at m/z 717 on stigma depleted much quicker than flavonoid species. Therefore, if the
number of microscans averaged for ions at m/z 717 are the same as the number of those for
flavonoid species, chemical distribution information of m/z 717 becomes indistinguishable
from the background. Figure S2 shows chemically-selective images from Arabidopsis
190
thaliana flower by using ‘selective spectra averaging’ method. When comparing chemically
selective images for ions at m/z 717 (Figure 7 and Figure S2), it is obvious that clearer
background was generated by selective spectra averaging method.
Jurchen, J.C., Rubakhin, S.S. and Sweedler, J.V. (2005) MALDI-MS Imaging of Features
Smaller than the Size of the Laser Beam. Journal of the American Society for Mass
Spectrometry, 16, 1654-1659.
191
Figure Captions
Figure S-1. Chemically selective images of ions at m/z 194, 210, and 226 and GALDI
mass spectrum of chloroform-dipped area. The mass spectrum was averaged
from 10 scanning points.
Figure S-2. Chemically-selective images of Arabidopsis flowers. Data were collected by
‘selective averaging’ method described in experimental procedure section.
The step size for data collection was set to 50 µm for both x and y directions.
Dimension of the images is 4700 µm high × 6700 µm wide. Chemical
compositions of corresponding m/z values were listed in Table 1. All images
were processed with absolute intensity values except the three mass images
for m/z 421, 451, and 978, which were processed with normalized intensities.
The maximum intensity value is listed under each image. For the image of
ions at m/z 717, the first two microscans were averaged for image presentation.
For the rest of the images, the first five microscans were averaged.
192
m/z 194 (-16) m/z 210 m/z 226 (+16) leaf1_30s_bottom_TI #6937-6946 RT: 97.53-97.67 AV: 10 NL: 2.96E4T: ITMS - c MALDI Full ms [150.00-1000.00]
150 200 250 300 350 400 450 500 550 600 650 700 750 800m/z
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Rel
ativ
e A
bund
ance
210.20
226.26194.33
255.40169.45
281.43241.36
473.42 517.42301.35 545.44445.38559.39403.34325.30 373.32 429.33 587.40
615.40 643.37 669.35 689.29 721.31 771.39
Figure S-1.
193
Figure S-2.
194
ACKNOWLEDGEMENT
First of all, I would like to thank my advisor, Professor Edward S. Yeung for the
generous support through my graduate works. I really enjoyed working with him and I have
learned a lot in all aspects. I sincerely admire him for his enthusiasm, dedication, and insight
to research. He is a great mentor and scientist.
I am also grateful to my POS committee members, Dr. Robert Houk, Dr. Victor Lin,
Dr. Basil Nikolau, Dr. Srdija Jeftinija, and Dr. Hans Stauffer for their precious advices and
supports.
I want thank all current and past Dr. Yeung group members for their help, advices,
and friendship: Becky Li, Melissa Zhang, Aoshuang Xu, Wei Sun, Wenjun Xie, Dr. Ning
Fang, Dr. Jiyoung Lee, Guoxin Lu, Slavica Isailovic, Dr. Frank Li, Dr. Dragan Isailovic, Dr.
Hungwing Li, Dr. Bob Hsin, and Dr. Seongho Kang. I specially thank to the mass
spectrometry project team, Dr. Hui Zhang and Dr. Chanan Sluszny. I really enjoyed working
with them. Besides our group members, my coworkers, Dr. Zhihong Song, Dr. Wenxu Zhou,
Dr. Libuse Brachova, and Dr. Ksenia Jeftinija deserve thanks for their direct assistance to my
research. In addition, I thank to my friends, Jeongyun Choi for helping Matlab programming
and Dr. Insik Jeon for precious comments.
I would like to thank all my friends for their support and happy moments we’ve
shared in Ames. Finally, I want to express my gratitude and love to my parents and all
families for their love, endless support, advices and… everything. Specially thanks to my
wife, Sang-Hee Shim for her support and love. You complete me.